You are on page 1of 34

Annual Reviews in Control 45 (2018) 118–151

Contents lists available at ScienceDirect

Annual Reviews in Control


journal homepage: www.elsevier.com/locate/arcontrol

Microgrid control: A comprehensive survey T


Sachidananda Sen , Vishal Kumar

Department of Electrical Engineering, Indian Institute of Technology Roorkee, Uttarakhand 247 667, India

ARTICLE INFO ABSTRACT

Keywords: Present researches in power and energy fraternity are driven towards the realization of smart grid (SG) tech-
Droop control nologies. Microgrids (MGs) being regarded as “elementary units” of SG, has undergone rigorous research for
Energy management system more than one and a half decade now. It provides an integration platform for microsources (MSs), loads, storage
Microgrid devices and power electronics (PE) converters at demand premises forming a system of systems i.e. SoSs ar-
Power sharing
chitecture. MGs mostly operated in grid-tied mode but during an emergency, are capable of standalone opera-
Smart grid
Voltage and frequency control
tion. Stability and operational control during both the modes are of utmost concern. Involvement of several
control functionalities viz. normalcy of voltage and frequency, optimal power sharing, islanding detection etc.,
had been adhered in the literature. In spite of that, the real world implementation has not been of significant
extent, that too even is far from having ideal MG characteristics, eligible for commercial usage. These technical
obstacles need to be identified and dealt with. This paper provides a comprehensive survey of different control
aspects of MGs, broadly classified under four control strategies: centralized, decentralized, distributed and
hierarchical frameworks. Each scheme is reviewed in detail w.r.t the principles behind, their applicability and
performances. It also identifies several research gaps and future trends therein. The technical barriers for real
scale application and their solutions are also briefly discussed. And finally, a discussion on different integrated
technologies for MGs, to realize SG features is also presented.

1. Introduction Lehn, 2005).


After the deregulated restructuring of power system (PS), penetra-
The microgrids (MGs) paradigm are not only economical, resilient tion of DERs is increasing at steady pace. Though it brings lots of
and reliable but also, provides environmental benefits as compared to benefits, it is also accompanied by several challenges. Firstly, the LV
the existing utility networks because of the use of renewable energy networks are getting converted from passive to active. Secondly, in-
resources (RESs). At 2.5% annual load growth (Annaswamy, 2013), a crease in complexity due to the use of PE converters. Thirdly, DERs with
huge unmatched gap arises in between demand and conventional different generation attributes i.e. hybrid, ac and dc create problems in
power generation. Also, transmitting power to remote demand sites voltage and frequency regulation. Lastly, two possible modes of op-
incurs huge losses. The MGs with features of quick installation and at eration i.e. grid-tied and islanded modes. These difficulties related to
demand proximity makes it a viable solution for the present energy- MG infrastructure, calls for a proper control architecture so that smart
related problems. The RESs viz. intermittent sources (solar photovoltaic grid features can be realized (Lamnabhi-Lagarrigue, 2017; Majumder,
(PV) cells and wind turbines), microturbines, fuel cells along with diesel 2010).
generators are mostly used in the MGs (Hatziargyriou, 2014). As several control aspects are involved in MG, the literature avail-
The intermittency and different generation characteristics require able is quite extensive. Different functionalities viz. droop control,
the microsources (MSs) to be placed on low voltage (LV) distribution voltage and frequency regulation, proportional active and reactive
feeders via storage devices (flywheel, battery and super capacitors) and power sharing, energy management system (EMS), MG optimization
power electronics interfacing (PEI) (mostly dc-ac or ac-dc-ac con- and multi-MGs interaction etc. being the fundamental and important
verters). Combination of each of these is also known as distributed issues for MG operation, are given the utmost priority. Because of MGs’
energy resources (DERs). Along with various DERs and loads (con- system of systems (SoSs) architecture these control requirements can be
trollable, shiftable and critical), as well as a specified control archi- fulfilled by using any established control system design strategies i.e.
tecture, forms the complete MG framework (Katiraei, Iravani, & centralized, decentralized, distributed and hierarchical frameworks.


Corresponding author.
E-mail addresses: sachi.dee2015@iitr.ac.in (S. Sen), vksaxfee@iitr.ac.in (V. Kumar).

https://doi.org/10.1016/j.arcontrol.2018.04.012
Received 14 March 2018; Received in revised form 24 April 2018; Accepted 24 April 2018
Available online 07 June 2018
1367-5788/ © 2018 Elsevier Ltd. All rights reserved.
S. Sen, V. Kumar Annual Reviews in Control 45 (2018) 118–151

Later, the smart grid (SG) functionalities can be built upon these MGs
control infrastructure (Martin-Martinez, Sánchez-Miralles, &
Rivier, 2016).
Though a few review literature on MG control are available, they
are not so comprehensive. A concise discussion on central controllers
for MGs are presented in Dragicevic, Lu, Vasquez, and Guerrero (2016),
Kaur, Kaushal, and Basak (2015). Similarly, Eid, Rahim, Selvaraj, and
Khateb (2016), Guerrero, Chandorkar, Lee, and Loh (2013), gives a
brief review on decentralized control techniques. A few distributed
control schemes were discussed in Yazdanian and Mehrizi-Sani (2014).
While in Bidram and Davoudi (2012), Canizares (2014), some appli-
cations on hierarchical control schemes are discussed. Some review
works are also available on MGs as a specific control function viz.
power sharing (Han et al., 2016), voltage and frequency control
(Malik, Ai, Sun, Zhengqi, & Shupeng, 2016). Most of these review lit-
erature discusses the principles behind these control strategies, how-
ever, fails to document the reported works completed in this field. Some
works compiled only a specific functionality and a few presents on the
basis of MG type i.e. ac, dc and hybrid. As a systematic and detailed
state-of-the-art comprehensive survey is lacking, this article brings to- Fig. 1. Microgrid architecture.
gether under one cover, a plethora of available MG control strategies
considering various important operational functionalities, designed 3. MG control: classification and functional layers
using both classical and modern control principles. Thereafter, several
future trends and research challenges for MGs’ real scale implementa- Since its inception, several MG control techniques have been re-
tion are also suggested. Along with that, role of MGs and some in- ported in the literature. As discussed in the introduction section, MG
tegrating devices for realizing smart grid concept is also discussed. This control systems can be broadly divided into four types: centralized,
article would assist smart grid researchers to have knowledge of ex- decentralized, distributed and hierarchical framework. In this section, a
isting control foundations realized and thereafter, building smart grid further detail on functional layers under each head is presented. It is
features upon them. important to note that, there is no unique control strategy for a specific
The paper is structured as follows. Section 2 briefs about MG ar- MG, as it depends upon the MG model, its configuration, and compo-
chitecture and control requirements. Section 3 gives the level based nents used.
classification of MG control techniques. The MG control literature is The centralized control scheme is usually implemented for small
addressed in Section 4. Several research challenges are presented in MGs. State variables information from component level sensors are
Section 5, discussions on smart grid realization are covered in Section 6 collected at MGCC via a high speed communication channel. The MGCC
and Section 7 concludes the paper. is responsible for processing and decision making as per status of the
state variables. The control law or set points for each LCs are sent
2. Microgrid architecture and control requirements through the same two-way high-BW link. These LCs adjust their re-
sponse as per new command. For grid-tied MGs and multi-MGs systems,
MGs are strategic entities located in the LV distribution network. It there is another controller in the distribution system. Hence, the central
is connected to the grid supply point (GSP) or the MV/LV substation at control strategy can be further divided into a three-layer structure
a single bus called point of common coupling (PCC) through circuit based upon the number of components and area brought under control
breaker or static transfer switch. It is controlled to switch between two i.e. distribution system management (DSM), MGCC and LCs.
modes of MG operation. A central controller for the whole MG is placed The decentralized control method uses local measurements and
on LV side of GSP and it is known as microgrid central controller based upon pre-defined algorithms embedded in them, it decides action
(MGCC) as shown in Fig. 1. It takes care of the power flow between the at the component level. Considering the fact of SoSs architecture of
upstream utility network and MG, cost optimization of MG and deciding MGs, multi-agent system (MAS) is best way to implement decentralized
mode of operation and islanding detection. Various types of MSs and control method. This control scheme mainly has two-stage architecture,
storage units are connected to LV radial feeders through PEI and MS mostly named as upper/higher and lower level controllers.
controllers (MC). The loads are also distributed along the feeders via Decentralized control is usually applied to the high rated MGs with
load controllers (LCs) for individual or aggregated loads large number of components where the centralized control structured
(Hatziargyriou, 2014). would be of sluggish nature (due to collection and processing of huge
The MG control functional requirements for economical and reliable data at one location i.e. PCC). Though it is debatable that which one
working can be formulated as: between centralized and decentralized architecture is better, both the
methods have their own advantages and disadvantages depending upon
• Regulation of voltage and frequency (V-f) within limits; operational scenario within the MGs. Table 1 presents a comparative
• Efficient load sharing i.e. active and reactive power (P-Q) balance summary.
and proper communication among DERs; The distributed control architecture is an improvisation of decen-
• Seamless transition between two modes of operation i.e. proper tralized technique, where each LCs communicate with its neighboring
synchronization and early islanding detection; ones (through low-BW channels in a collaborative manner) to attain
• Economic dispatch of the DERs units by optimization to reduce MG some advantages of centralized architecture meant for the whole MG. It
running costs; means, each device is controlled based upon local measurements by
• Power flow control between MG and PS network i.e. an energy taking account of neighbors’ response. It not only preserves the au-
management system (EMS). tonomy of each controller but also, simultaneously achieves the global
optimization by dividing a large problem into several sub-problems.
MAS is the most sophisticated tool for implementing this scheme where

119
S. Sen, V. Kumar Annual Reviews in Control 45 (2018) 118–151

Table 1 4.1. Droop control


Centralized vs decentralized control.
Factors Centralized control Decentralized control 4.1.1. Centralized control technique
Droop control of MGs is a concept where the conventional power
DG ownership Single owner Multiple owners system (CPS) is imitated i.e. when demand for active power increases,
Goals A clear single task Uncertain tasks
the frequency falls and vice-versa. Similar relations are observed be-
Operating personnel Available Not available
Flexibility Less More
tween the reactive power and voltage. Now, as CPS consists of huge
Market participation Complex algorithms, all Simple algorithms, Some synchronous machines with large inertia, this functionality is auto-
units collaborate competitive matically accomplished. But, for MGs mostly inverter interfaced sources
Plug-and-play Not possible Possible are used and hence, need for separate droop control is inevitable.
Optimality Optimal solutions Suboptimal usually
Although droop control is a component level functionality, centralized
Expansion Troublesome Easy
Communication needs High Low control technique has also been applied in some of the available works.
Setup cost High Low A centralized droop control scheme for parallel operation of DERs in-
More critical operation Possible Not possible verters under both grid-tied and islanded operation is proposed in
Failure occurs If one faulty point Many points faulty
Tan, Peng, So, Chu, and Chun (2012). In grid-tied mode, the controller
operates in current control mode, while in islanding operation it works
as voltage control mode. A model predictive control (MPC) strategy is
each controller is considered as an intelligent agent. It also has a three-
used and the complete problem is segregated into two sub-problems as
layered structure: droop/lower, secondary/intermediate and auxiliary/
transient and steady state. This leads to faster parallel computations
upper. As this control scheme incorporates the advantages of both
and reduced response time. But, as Kalman filters were integrated for
central and decentral architecture, it is the most exploited one in lit-
generating the controller references, overall design complexity is quite
erature.
high.
Hierarchical control structure is based upon the difference in time
In Yao et al. (2013), a MG having resistive feeders is used to im-
scales of various control requirements. It is derived from ANSI/ISA-95
plement the reverse droop control scheme i.e. P-V and Q-f based. The
(International Society of Automation) which is a standard meant for
MGCC is designed to control the parallel operation of inverters, along
interface or interoperability among different enterprise and control
with usage of active compensator/filters to eliminate the circulating
systems. It is also divided into three-level structure: primary, secondary
currents. Classical PI controllers were realized and three MG operating
and tertiary. The primary layer is meant for instantaneous control of the
modes of were studied. It not only accomplishes the task of droop but
MGs’ device level state variables viz. V-f regulation. It operates in
also, maintains system stability all operating modes. The control laws is
milliseconds and seconds time interval. The secondary control layer is
(1):
slower (within minutes) for eliminating steady-state deviations arising
from action of primary layer. The tertiary or auxiliary layer is the Vi = Vi* m (Pi Pi*) + Vt
slowest one (several mins to an hour) and manages power flow/ex- fi = f i* + n (Qi Qi*) + ft (1)
changes with the upstream utility grid. It implements the economic
dispatch and optimization of overall MG system. with deviation given as:
Distributed and hierarchical control strategies complement each
Vt = Kpv (Vgrid Vmg ) + Kiv (Vgrid Vmg ) dt
other. Both need same kind of communication and control infra-
structure. Fig. 2 shows complete MG control structure. ft = Kpf ( grid mg ) + Kif ( grid mg ) dt (2)

For parameter definition and values refer Yao et al. (2013).


4. Microgrid control
4.1.2. Decentralized control technique
This section presents a detailed survey of important control objec-
tives for MGs viz. droop control, V-f control, P-Q sharing, EMS and This control scheme is mostly exploited for implementing the droop
several other miscellaneous functions accomplished using aforemen- control for MGs. Early works in Guerrero, Matas, Vicuna, Castilla, and
tioned four categories of control architectures. Each of the control Miret (2007) gives a novel decentralized strategy for parallel working
functionalities is discussed in a structured comparative manner, with of inverters in an islanded MG. The LV feeders were considered to be
some critical comments on benefits and limitations of each of the resistive and hence, conventional droop control is not applicable. The
control techniques. controller performs the V-f control as well as P-Q power sharing using
three nested loops: (1) output voltage regulating inner loop; (2) output
resistive impedance loop; and (3) outer loop for P-Q sharing. It suggests

Fig. 2. Classification of microgrid control techniques and functional layer structure.

120
S. Sen, V. Kumar Annual Reviews in Control 45 (2018) 118–151

a P-V droop and Q-ω boost scheme using two PD controllers to satisfy all MSs becomes equal in synchronization to frequency, the optimal
the parallel inverters operational requirements. This controller is able economic dispatch of MSs is achieved. Hence, this scheme also helps to
to mitigate harmonic imbalance and provide increased damping, hence, optimize the total operating cost, that too without the use of any
is widely accepted. The droop equations are given as: communication. The equations are given as:
dP fi = fmax mIC ·ICi (Pi )
V = V* nP nd
dt fmax fmin Ci (Pi )
dQ mIC = with ICi (Pi ) =
= * mP md fmax Pi (4)
dt (3)
Similarly, in Vu et al. (2017), the droop function for dc MG is rea-
Thereafter, a decentralized droop strategy for a LV dc MG is pro-
lized using closed-loop reference model (CRM) adaptive control tech-
posed in Khorsandi, Ashourloo, and Mokhtari (2014). It takes into ac-
nique. A time-varying model is derived thereafter, a normalization
count the effects of unequal feeder resistances and state-of-charging
method and projection algorithm, with CRM is used to achieve si-
(SoC) of the battery energy storage systems (BESSs) in presence of solar
multaneous current sharing, voltage stability. It also suggests an
PVs based sources. Three modes of operation and fault conditions are
adaptive gain scheduling procedure. But, its implementation is limited
considered, to represent perfect power sharing among the DERs. Clas-
to dc MGs only. Recently, a new GPS based technique for synchroni-
sical PI controllers were used to realize the scheme, various tests sug-
zation of DERs within islanded MGs is proposed Golsorkhi, Lu, and
gest a smooth operation of the MG with insignificant error in power
Guerrero (2017). Rotating reference frames of MSs controllers are
sharing. However, it didn't discuss its applicability towards ac MGs. In
synchronized, which later, were used for the V-I droop method. Along
Gavriluta et al. (2014), a customized droop scheme combining dc-bus
with it, an adaptive Q-f droop scheme is realized as a backup in case
signaling (DBS) and droop concepts for multi-terminal dc (MTDC)
there is an interruption in receiving GPS signals. Though this scheme
networks is presented. Five operating bands viz. normal, safety (high
enhances the controller reliability, however, its implementation com-
and low) and critical (high and low) similar to normal, alert and
plexity and real-scale feasibility are the main concerns.
emergency of power system is considered. The SoC of storage units and
Another adaptive droop scheme for LV dc MGs based on super-
dc-bus voltage were also included to create a two-variable droop sur-
imposed frequency accompanied with a virtual resistor is proposed in
face for controlling BESSs. It protects the batteries from overcharge and
Peyghami, Mokhtari, and Blaabjerg (2017). Better load sharing and
deep-discharge.
voltage regulation than conventional droop are achieved using locally
The economic operation of MG, which is generally realized in sec-
available measurements to adapt the droop gains by introducing an ac
ondary level control can also be achieved by the primary level adaptive
power at both primary and secondary levels. This avoided the need of
droop control strategy as discussed in Oureilidis and Demoulias (2015).
any communication infrastructure and hence cost effective. The droop
Here, impedance of the MG as measured by each DERs are used to
characteristic is described by (5). Later, the controller parameters are
adjust the droop coefficients. DERs nearby the loads provides most of
chosen based upon stability analysis of a developed small signal model.
the power than farther ones. Hence, as the current is flowing through
The main concerns of the proposed method is injection of ac power into
least impedance path, transmission power loss is reduced. It also de-
the dc MG, which can deteriorate its power quality and cause stability
couples the active and reactive power (P-Q), therefore, avoids con-
problem.
trolling of the DERs virtual impedance. Effectiveness of proposed con-
troller is tested for both radial and looped MG topologies. This scheme dpk G (s ) Qk
* = V*
V ok Rdk Iok + Iok
provides scope for operational optimization of MGs. Iok
A V-I droop characteristic is given in Golsorkhi and Lu (2015). By dpk G (s ) Qk
this method, the nonlinear power sharing problem is converted to a * = V*
V ok R^dk Iok where R^dk = Rdk +
Iok (5)
simple piecewise linear current sharing. It not only enhances the
damping but also, accuracy of DERs during overloading operations. More detains are given in Peyghami et al. (2017). A novel f-P/Q
Same authors extended this work to improve the power quality of MGs based droop control for synchronization and power balance of DERs in
in Golsorkhi and Lu (2016). Here, MPC combined with V-I droop works cascaded-type MGs is presented in Sun et al. (2018). It provides the
in coordination to prevent active power overloading, mitigate voltage advantage of its applicability to both resistive-inductive (RL) and re-
unbalances by adjusting negative sequence output impedance of MSs sistive-capacitive (RC) type loads and also accomplishes the task of bus
and limiting circulating currents. It tunes the controller parameters level voltage regulation. However, application of proposed droop
heuristically to reduce the effects of disturbances. However, usage of scheme on parallel or interconnected MGs has not been explored.
GPS for synchronization purposes enhances its overall complexity and
brings reliability issues. 4.1.3. Distributed control technique
With inclusion of devices having different characteristics within the First distributed control scheme for secondary level droop control is
MGs, makes the developed droop controllers’ usage inappropriate. This presented in Shafiee, Vasquez, and Guerrero (2012). It attains proper
calls for certain modifications in the droop control strategies. In sharing of reactive power among the DERs. PI control is used to reduce
Bhattacharya and Mishra (2016), a MG with PVs and electric vehicles steady-state deviations in V-f at the secondary level. Thereafter, the
(EVs) as MSs are considered to implement a modified droop control effect of communication delays was studied in comparison to a cen-
scheme. During grid-tied mode PV works in MPPT and EVs are charging tralized secondary method. Benefits of this networked controller it that,
but, in islanded mode, constant PQ control is used at ac side for the operation of the whole system doesn't fail for the failure of a single
seamless mode transfer without any control switching. For dc side, a device and also gives better stability performance.
double quadrant droop scheme is proposed for charging to equalize the A distributed average PI (DAPI) droop scheme for power sharing is
SoCs during vehicle to grid (V2G) mode. This controller has the features given in Simpson-Porco, Shafiee, Bullo, Shafiee, and Guerrero (2013).
for implementation on a hybrid MGs. Similarly, an adaptive droop The controller developed is robust against small load variations and
strategy for dc MGs is presented in Sahoo, Mishra, and Padhy (2016). maintains system frequency within allowable range by utilizing La-
Solar PVs, fuel cells (SOFC) and BESSs are considered, the charging and placian averaging method. It also gives existence conditions for stable
discharging of BESSs is managed using bidirectional adaptive strategy operating points. In Simpson-Porco et al. (2015), same authors extend
for proper power sharing. Q. Xu, Wang, Zhang, Wen, and Xiao (2016) this work further and proposes a generalized distributed architecture
gives a novel droop scheme based upon frequency and incremental cost for V-f control at secondary level of an autonomous MG. This controller
(f/IC) of MSs in an autonomous MG operation. At steady state, as ICs of helps to make intuitive tradeoffs between two conflicting goals of

121
S. Sen, V. Kumar Annual Reviews in Control 45 (2018) 118–151

voltage regulation and Q sharing. The DAPI based frequency controller control scheme that balances charging and discharging cycles of dis-
is given in (6). tributed BESSs is proposed. While charging, as frequency of MG in-
creases, the droop coefficients are adjusted proportional to SoCs of
= * mi Pi +
i i
n batteries and during discharging, these coefficients are set as inversely
d i proportional to SoC as in this case system frequency decreases. This
ki = ( i *) aij ( i j)
dt j =1 (6) presents an additional degree-of-freedom (DOF), but its efficacy is yet
to be verified. A DAPI based droop is used to study transient perfor-
An adaptive droop control scheme is proposed in Nasirian, Davoudi, mance of a multi-inverter MG (Tegling, Andreasson, Simpson-Porco, &
and Lewis (2014), where virtual impedances of DERs are controlled by Sandberg, 2016). The power losses due to resistive line, while reg-
current regulators to implement proportional load sharing. A PI based ulating the frequency during load or generation variation, acts as small
voltage regulator is also introduced to compensate for any voltage disturbances of persistent nature. Input-output H2 norm is used to
droop due to droop controller. The voltage regulator uses average quantify the transient power losses. A comparison between the pro-
voltage value of MG estimated by a voltage observer by using dynamic posed DAPI and conventional secondary level droop is also done. Re-
cooperative averaging information obtained from neighboring con- sults shows losses can be reduced by optimal tuning of DAPI and only
verters. Advantages of this controller lie in the fact that, a sparse cyber low-gain distributed feedbacks can optimize MGs’ transients.
network is used for data exchanges which also aids implementation of
the cooperative algorithm. Similarly, an advanced droop scheme for dc 4.1.4. Hierarchical control technique
MGs are proposed in Maknouninejad, Qu, Lewis, and Davoudi (2014). Applicability of hierarchical control technique do not have much
Here, a nonlinear optimal controller for voltage regulation and power scope as droop controller hence, its usage is very limited. A two stage
sharing is presented. The control strategy needs full information of MG droop controller for current sharing among dc-dc converters in a dc MG
hence, it is converted to a linear near-optimal control problem. The is given in Srinivasan and Kwasinski (2015). The primary layer is
performance index is given in (7). It is found to be robust, stable and constructed using passivity-based controller that yields virtual resistors
have better performance than usual droop schemes even for inter- at converter output. At secondary layer, voltage regulator is added
mittent communication links. Under a special case, the proposed con- using an integral control, for compensating the steady-state voltage
troller can work as conventional droop control strategy hence, is more deviations, and attaining asymptotic stability at local equilibrium
reliable. However, it is limited to dc MGs only. points. Although this technique is able to replicate the static droop
T
characteristics without any communication links, its usage is confined
J= [Pgi (T ) Pg*i]2 + [ki2 (Pgi Pg*i)2 + ui2] dt to dc MGs only and gives a sluggish response. Table 2 briefly sum-
i i
marizes the droop control techniques.
t0 (7)

Multi-functional control scheme for coordinated operations of VSIs 4.2. Voltage and frequency control
in MGs is proposed in Chang and Zhang (2015). It accomplishes pri-
mary droop and secondary control for power sharing and to synchro- 4.2.1. Centralized control technique
nize the frequency subject to phase angle and voltage constraints. It is Use of droop controller for proper load sharing among the DERs
applicable to both radial and ring MGs with resistive/inductive lines. leads to persistent small and large fluctuation in system voltage and
Linear matrix inequalities (LMIs) concept is used for controller design frequency (V-f) and this might result into equipment failure, distorted
which provides smooth operation, robustness against small dis- power quality, stability problems etc. Hence, maintaining the V-f within
turbances. Most appealing feature of this controller is its applicability to operating range is another important control objective for MGs and this
meshed lossy MGs with uneven R/X distribution lines. objective can be fulfilled by aforementioned control schemes. In this
A distributed droop control for interconnected multiple dc MGs regard, early work by Akbari, Golkar, and Tafreshi (2011) proposes a
forming a networked MG is proposed in Weaver, Robinett, Parker, and centralized voltage control method with both ac and dc sources under
Wilson (2015). It uses the local BESSs for each MSs to actuate the droop grid-tied mode. The PI based controllers were designed for various PE
characteristic. The MSs voltages are matched to bus voltage by peri- converters, MSs and storage devices to have a centralized coordination
odically updating the duty cycle of the dc-dc converters at asynchro- based control system. This controller has robust capabilities in restoring
nous intervals through a feed-forward method. Results suggest a re- voltage stability by taking account of generation uncertainty of MSs.
duction in number of storage units’ requirements for maintaining Thereafter, in Tabrizi, Radman, and Tamersi (2012), voltage profile of
transient stability. Also, usage of Hamiltonian surface shaping (HSS) MG under two scenarios: the transition from grid-tied to islanded mode
makes tuning of PI controller gains easy to eliminate the steady state and for a 3-phase short circuit fault during an islanding operation is
error of the overall system. maintained using MG voltage controller (MGVC). The MGCC based
A multi-objective cooperative control strategy is proposed in lead-lag controller is designed to function as MGVC. It also supports the
Lai, Zhou, Lu, Yu, and Hu (2016). It converts the V-f regulation into existing reactive power controllers in mitigating voltage deviation
pinned synchronize control strategy and P-Q sharing to an average across all feeders and loads. Simulation results show improved voltage
consensus based problem. Which later, provides the reference set points profile for pre- and post-islanding conditions. Fig. 3 depicts structure of
for the droop scheme. It uses communication network with switching MGCC in a distribution network.
characteristics and can handle non-uniform time-varying delays. Sta- An adaptive controller for under frequency load shedding is pro-
bility analysis is done using Lyapunov–Krasovskii method to obtain posed in Karimi, Wall, Mokhlis, and Terzija (2017). It includes a state
sufficient conditions for network links and upper limit of communica- estimator, as MGs have a limited number of measurement units at some
tion delays. These criteria make its reliable usage under time-varying buses. It helps to compute the power imbalances using power flow
delays. measurements and rate of change of frequency (ROCOF).
In Avila and Chu (2016), a new droop control scheme for P-f and Q- The controller has two operating modes: (1) event based meant
V for primary level of isolated MG is presented. At secondary level, this during islanding of MG or tripping of a generator. (2) response based
simultaneously accomplishes the task of accurate P-Q sharing. The during sudden rise in load demand. It is divided into four modules to
controller is based upon distributed pinning MAS. The source agents are calculate event and frequency, state estimator and load shedding
chosen as pinning agent i.e. virtual leader while, load agents as reaction modules. This modular design can facilitate easy modification of con-
mechanism. This control scheme needs low BW communication chan- troller task in future. However, usage of distributed state estimator
nels. In Gkavanoudis, Oureilidis, & Demoulias (2016), adaptive droop (DSE) makes it quite sluggish.

122
Table 2
S. Sen, V. Kumar

Summary of droop control techniques for microgrids.


Design technique Authors Control principle Contribution Comments (Benefits/limitations)

Centralized control (Tan et al., 2012) Model Predictive Control (MPC) Complete problem is divided into two subsystems: transient, Faster transient response, requires complex parameter
steady-state. estimations.
(Yao et al., 2013) Proportional-Integral (PI) control Reverse droop (P-V and Q-f) is implemented as the network is Improves stability and eliminates circulating currents.
considered to be resistive.
Decentralized control (Guerrero et al., 2007) Proportional-Derivative (PD) Novel three nested loops for P-V droop and Q-f boost. Provides more damping, mitigates harmonics, good dynamic
control response.
(Khorsandi et al., 2014) PI control Effects of unequal feeder resistances and SoC of BESSs were Represents perfect power sharing, but restricted to dc MGs
included. only.
(Gavriluta et al., 2014) Custumized droop control using A new droop concepts for multi-terminal dc (MTDC) SoCs included, batteries are protected from overcharge and
DBS networks. deep-discharge.
(Oureilidis & Demoulias, 2015) Adaptive droop control Droop coefficients are adjusted using measured impedance of Transmission power loss is reduced, independent of network
MG by each of the DERs. topology.
(Golsorkhi & Lu, 2015; M.S. 2016) MPC combined with V-I droop Nonlinear power sharing is converted to simple piecewise Parameters are tuned heuristically, effects of system
linear current sharing with enhanced power quality. disturbances are reduce, but, GPS increases complexity.
(Bhattacharya & Mishra, 2016) Modified droop control Double quadrant scheme to include EVs within MGs. Suitable for hybrid ac/dc MGs, vehicle to grid (V2G) mode is
possible.
(Sahoo et al., 2016) Bidirectional adaptive droop Integrates various RESs (PVs and SOFC) and BESS. Manages proper power sharing, applicable for dc MGs only.
control
(Q. Xu et al., 2016) Decentralized droop control Novel scheme based on frequency and incremental cost (f/IC) Optimizes total operating cost, ED without any
of MSs. communication.
(Vu et al., 2017) Closed-loop reference model (CRM) Time varying model for dc MGs, normalization and projection Achieves simultaneous current sharing, voltage stability and
adaptive control algorithm for CRM. adaptive gain scheduling.

123
(Golsorkhi et al., 2017) V-I and adaptive Q-f droop GPS based technique for synchronization of DERs. Enhances controller reliability, real-scale feasibility is difficult.
(Peyghami et al., 2017) Adaptive control Droop control for LV dc MGs based on superimposed ac power Improved load sharing, voltage profile but, deteriorated its
frequency. power quality.
(Sun et al., 2018) f-P/Q droop control A novel droop control for synchronization and power balance Applicable to both RL and RC type loads, However, its usage on
of DERs in cascaded-type MGs. Parallel or interconnected MGs is not explored.
Distributed control (Shafiee et al., 2012) PI control First distributed control scheme for secondary level droop Better stability performance, higher resiliency, reduce steady
control. state deviations in V-f.
(Simpson-Porco et al., 2013; Simpson-Porco Distributed average PI (DAPI) Robust droop scheme against load variations, maintains Makes intuitive tradeoffs between two conflicting goals i.e.
et al., 2015) control frequency using Laplacian averaging method. voltage regulation and Q sharing.
(Nasirian et al., 2014) Adaptive droop and PI control Virtual impedances of DERs are controlled by current Sparse cyber network is used for data exchanges, meant for dc
regulators, voltage regulator compensates drop. MG, need voltage observer.
(Maknouninejad et al., 2014) Optimal control A nonlinear optimal controller for voltage regulation, power Although robust, however, only near-optimal control solution
sharing along with droop control. is obtained.
(Chang & Zhang, 2015) Linear matrix inequalities (LMIs) Primary droop and secondary control for power sharing, also Robustness against small system level disturbances, applied to
synchronize the frequency. meshed lossy MGs with uneven R/X lines.
(Weaver et al., 2015) PI control with HSS based tuning A distributed droop control for interconnected multiple dc Reduction in number of storage units, eliminate steady state.
MGs.
(Lai et al., 2016) Pinned average consensus based A multi-objective cooperative control strategy for ac MGs Reliable for time-varying non-uniform delays, improved
control based on droop technique. stability.
(Avila & Chu, 2016) Pinning multi-agent system (MAS) New droop control scheme for P-f and Q V for primary Simultaneously achieves P-Q sharing, needs low BW
level. communication links.
(Gkavanoudis et al., 2016) Adaptive droop control Balances the charging and discharging cycles of distributed Adjusts droop coefficients proportional to the SoCs of batteries,
BESSs. more DOF.
(Tegling et al., 2016) DAPI and H2 norm based optimal A DAPI based droop is used to study transient performance of Losses are reduced by optimal tuning, optimized transient
control a multi-inverter MG. dynamics.
Hierarchical control (Srinivasan & Kwasinski, 2015) Passivity and integral based A two stage droop controller for current sharing among dc-dc Attains asymptotic stability for local equilibrium points, virtual
controller converters in dc MGs. resistors gives the static droop.
Annual Reviews in Control 45 (2018) 118–151
S. Sen, V. Kumar Annual Reviews in Control 45 (2018) 118–151

with the controller, hence, is found to be more accurate.


Thereafter, Etemadi, Davison, and Iravani (2014) presents a de-
centralized robust servomechanism based controller for an autonomous
MG. Here, the frequency is controlled by internal oscillators using GPS,
while a centralized power management system (PMS) decides the vol-
tage and power set point of DERs. Robustness is attained by eigenvalue
and real stability radius studies. HIL test results suggest fast tracking of
set points zero steady state error, robustness under loads variations.
Similarly, a robust Linear Quadratic Regulator (LQR) based controller
for V-f regulation is presented in Hossain, Lu, Mahmud, and Aziz
(2014). Multi-inverter nonlinear MG model with network uncertainties
is used as test system. The controller is tested for intentional islanding
and severe three-phase faults, which satisfactorily provides the required
level of stability and robust transfer from grid-tied to islanded mode.
Limitation of this controller is that its stability region is confined to a
Fig. 3. MGCC structural layers (Tabrizi et al., 2012). small post-fault operating point.
In Huang, Liu, Xiao, and Moursi (2015), a novel two-level voltage
4.2.2. Decentralized control technique control for dc MGs is proposed. The primary layer is based on the droop
In Ahn and Peng (2013), a decentralized scheme for voltage control scheme to manage load sharing. The secondary layer employs an
of small LV military MGs is proposed. The model-free control archi- average voltage sharing (AVS) using a bi-proper anti-wind-up method
tecture is divided into two levels: 1) lower level controller, consists of a and PI control for parallel operation of VSCs. Then, a pilot bus reg-
phase locked loop (PLL) and an inverter controller, to regulate terminal ulation is integrated to achieve precise voltage regulation by adjusting
voltage and output power. (2) the higher level controller, equipped converter reference points. It enhances system reliability and avoids
with cost function to minimize the distribution power loss. It also helps saturation of controller efforts. In Sadabadi, Karimi, and Karimi (2015),
to realize the plug-and-play capacity. Advantage of using simple in- voltage control of MGs with a radial topology is proposed. A novel LMI
verter model makes it highly efficient, but, as voltage levels for the load based problem formulation of fixed order control of RESs subject to
buses were not monitored, its applicability on large MGs is limited. polytopic uncertainties done. The controller structure, its parameters
Fig. 4 shows the proposed two level control structure. In Luo, Lai, Loo, were simultaneously designed using convex optimization technique. A
Tse, and Ruan (2013), a triple-layer droop control scheme for parallel linear time-invariant (LTI) state space model is derived to validate the
inverters is presented. The 1st PID droop controller works as conven- voltage tracking and robust behaviour under load variations of the
tional P-f and Q-V scheme while, 2nd one sets references for 1st by proposed controller. Similarly, Riverso, Sarzo, and Ferrari-Trecate
monitoring dc link voltage, and 3rd is similar to 1st one, but acts on dc- (2015), emphasizes on realizing plug-and-play (PnP) capability of MSs
dc converter interfaced MSs. It works in both modes and achieves is- by an LMI based problem. Usage of quasi-stationary line (QSL) ap-
landing without any communication. But, this multi-level approach proximations of distribution network helps in achieving smooth inter-
causes a delayed control action. actions among various MSs is also discussed. This design feature gives a
A Linear Matrix Inequality (LMI) technique based robust droop benefit to include neutral interactions for meshed MG networks without
scheme for stable operation of MG is proposed in Trung (2013). This using a global MG model. On a similar note, Tucci, Riverso, Vasquez,
controller provides a simple optimal power sharing techniques along Guerrero, and Ferrari-Trecate (2016) performs testing and validation of
with guaranteed zero steady-state frequency deviation and robust vol- a voltage controller.
tage stability under the influence of dynamic load variations. The ro- A voltage regulation method for islanded MG is presented in
bust capability to regulate the MGs’ V-f under bounded variations, gives Toub, Aniba, Maaroufi, and Robinet (2015). The decentralized power
this scheme an edge over earlier controller designs. A decentralized flow control is realized by Hamiltonian Surface Shaping (HSS) and
multi-agent (DMAS) technique is used for frequency control of isolated classical PI method. The controller has both PnP as well as scalability
MG in Liu et al. (2014). It accomplishes cooperative frequency control features and the control strategy shifts the poles to further left of the
using two levels: primary layer by agents of BESS and secondary layer imaginary axis hence, increases the stability. In
with MS agents. Communication constraints were also imposed, where Lizhen, Xiaohong, Zhengzhe, Zuohao, and Hu (2016), the local DERs
each agent can communicate only with its neighboring agents. Usage of controllers are divided into primary and secondary levels for compen-
an optimized Average Consensus Algorithm (ACA) is done to provide sation of voltage unbalances. The secondary layer collect data from all
information of global power deficiency to the DMAS. This algorithm DERs and generate control law for primary level control by an aver-
makes the parameter estimation process a cooperative arrangement aging technique. A similar three-level quasi-hierarchical structure is
given in Zhang, Xin, Wang, and Gan (2016). Here, a droop control for
both voltage- and current-controlled VSCs that interfaces various MSs
within the MG by implementing a simple local PI based structure.
In Sadabadi, Shafiee, and Karimi (2017), the PnP ability of MSs is
represented as a LTI system with polytopic uncertainty. QSL based
model and a robust control strategy based on the convex optimization
method is proposed. Results suggesting guaranteed overall stability,
desired voltage tracking along with robustness to topological changes
without any requirement of retuning of controller parameters. How-
ever, it uses GPS based synchronization hence, failure rate can be high
when signal is interrupted. In Gu, Lou, Tan, and Yuan (2017), a sec-
ondary voltage controller is given using nonlinear state estimation
method. A large-signal dynamic model of MG is derived and using a
linear parameter varying (LPV), the state dynamics of each DERs are
estimated. Then, the controller is designed by LMI and pole placement
Fig. 4. Two level decentralized controller (Ahn & Peng, 2013). conditions. Fig. 5 shows block diagram of the proposed method. This

124
S. Sen, V. Kumar Annual Reviews in Control 45 (2018) 118–151

Later, in Schiffer, Seel, Raisch, and Sezi (2016), same authors extended
this work and proves the dependency of equilibrium points for closed
loop dynamics of V-Q on selection of control parameters. Although this
technique gives a performance gain against usual voltage droop
schemes and has applicability on arbitrary network topology, the as-
sumption of small phase angle at DERs output is not appropriate for
isolated MGs.
A distributed secondary control (DSC) for regulating power, V-f is
proposed in Shafiee et al. (2014). A wire-less periodic broadcast com-
munication using token or discrete data packets which is simulta-
neously coupled to robust control functionality is presented. The LCs
update the control law by acting on each data exchanges to achieve the
desired response hence, avoids data loss and time delay. Same authors
give a cooperative algorithm for MGs’ frequency synchronization in
Shafiee, Vasquez, Guerrero, Nasirian, and Davoudi (2015). It does not
need to measure the frequency, rather the average active power is es-
Fig. 5. Decentralized secondary voltage control scheme (Gu et al., 2017). timated using consensus method and later, is used as a correction term
to primary droop scheme to achieve the active power sharing and keep
the frequency around the nominal value. However, the control action is
schemes not only achieve average voltage restoration but also facil-
slow.
itates accurate reactive power sharing.
In Guo, Wen, Mao, and Song (2015), a secondary V-f control scheme
In Cook, Parker, Robinett, and Weaver (2017), the HSS based power
for finite-time voltage regulator to track reference values is given, along
flow control for dc MGs is proposed. Three control methods viz. optimal
with consensus-based frequency control scheme (which is decoupled
exergy destruction (OXD), droop control and the decentralized mode-
from voltage controller), subject to control input constraints for pro-
adaptive (DMA) were designed for voltage regulation. OXD method
portional power sharing. This design gives an edge over the prior ones
helps to reduce the energy loosed in VSCs and DMA minimizes usage of
in terms of speed of enforcing the control law without much loss of
storage units. All these control strategies use HSS. These OXD and DMA
stability. Related works are also presented in Chen and Feng (2015),
approaches provide a new control technique whose efficacy lies
where each control modules: voltage, frequency and power sharing are
somewhere in between droop and centralized strategy.
realized as primary i.e. LCs. Here, although control action is much
faster, its stability margins are lower. Later, a three level structure for
4.2.3. Distributed control technique frequency control and ED is proposed in Xin, Zhang, Wang, Gan, and
The very first work combining advantages of both central and de- Wong (2015). It uses an equal incremental cost function to restore the
central control schemes is given in Wu, Tang, Dragicevic, Vasquez, and frequency in allowable range. LPFs are designed to decouple the dy-
Guerrero (2013). A two layer coordinated control strategy for frequency namics and enhance stability. The Control law is given by (8):
control is proposed. Primary level has a master-slave architecture
where BESS acts as master, maintains V-f based on the frequency bus fi, des = fpri (Pi ) + fsec ( , fi, des ) + fter ( , Pi, i (Pi ))
signaling (FBS) and depending upon its SoC, the RESs acts as slave units with fpri (Pi ) = fmax di Pi
working on virtual inertial to generate maximum power. Secondary
(50 fi, meas )
layer constitutes a PI controller to mitigate the frequency deviations fsec =
(1 + T2 s )
arising from lower controller. It has a simple implementation and only
needs low cost communication channels. However, its accuracy and (fmax di Si i (Pi ) fpri )
fter =
speed of control action are not satisfactory as monitoring SoC of BESS is (1 + T3 s ) (8)
involved.
In Bidram, Davoudi, Lewis, and Guerrero (2013), a secondary level For more details refer Xin et al. (2015). In Cady, Domínguez-García,
voltage control is proposed based on cooperative MAS. Dynamics of and Hadjicostis (2015), a distributed control to regulate frequency and
inverters are modeled as nonlinear and unequal, then by using input- optimal dispatch of MGs with both synchronous and inverter interfaced
output feedback linearization, this is transformed into a framework of MSs is discussed. Here, linear iterative algorithms are integrated into
second order linear tracking problem to achieve synchronized opera- LCs to provide set points for DERs using local measurements only. The
tion of multiple DERs. It also provides desired speed of response by controller is event triggered i.e. reference points are updated at any
tuning of control parameters. In Bidram et al. (2013), same authors time instant (aperiodic) so that, during constant power requirements,
attempt this work for frequency control. Later, in control effort is zero. Later, same authors in Cady et al. (2015), pro-
Bidram, Davoudi, Lewis, and Ge (2014), it is extended to incorporate a poses a synchronization method for MSs. It consists of LCs with two
linear neural network architecture to compensate uncertainties due to algorithms: firstly, for a given load demand, determination of set points
unknown parameters of the DERs. Results suggest that overall sto- that makes the MG stable and in synchronism to the upstream grid.
chastic nature are well addressed, however, design complexity is in- Thereafter, converging the weighted steady state frequency error to
creased immensely. Similarly, Serban (2014), includes BESS for fre- zero by disturbance rejection using a ratio consensus algorithm and PI
quency control. It considers the dead-zone nonlinearity and uses PD and controllers. Both these algorithms help to attain a smooth MG operation
PID controllers, when Δf is within and beyond range respectively. The even under persistent small deviations in V-f.
primary uses a PD control action, as derivative term improves MGs’ A MAS based control structure for multi-inverter MGs is proposed in
frequency dynamics by providing a virtual inertia. Simple to implement Rahman, Mahmud, Oo, and Orchi (2015). It provides a stable load V-I
and faster transient response are the two benefits of this controller. during normal operation, under faulty conditions by asymptotic
A distributed voltage control scheme based on a consensus algo- tracking of the preset reference current signal. PI controllers are de-
rithm for an inductive MG is given in Schiffer, Seel, Raisch, and Sezi signed as agents and tuning is done by using Ziegler–Nichols method.
(2014). It simultaneously accomplishes the task of steady-state reactive Comparison of proposed scheme is done with existing conventional
power sharing using communication by graph theory. Results suggest it controllers without MAS, results suggest it has a better performance and
is successfully able to provide exponential stability at the local level. ease of parameter adjustments.

125
S. Sen, V. Kumar Annual Reviews in Control 45 (2018) 118–151

the selected number of MSs, and all MSs exchange information with
neighbouring ones. This helps to achieve P sharing and V-f restoration
at a speed having exponential convergence. This controller can update
its states by impulse jumps and in discrete time with time-varying de-
lays. Same authors in Lu, Yu, Lai, Guerrero, and Zhou (2017), propose a
DSC for V-f control during islanded operation. Here, an uncertain and
sparse communication link is used for data sharing. V-f restoration and
P-Q sharing are accomplished using two discrete-time DSC controllers
working on iterative learning methods. Stability and robustness con-
ditions were also derived using LMI, algebraic graph theory and the
Lyapunov stability theory. Similarly, in Lu et al. (2017), this work is
further extended to a multi-purpose secondary level, with two con-
trollers. First one provides finite time frequency regulation along with P
sharing. Later one is meant for voltage regulation and Q sharing at a
slower time scale. Hence, the functioning of these gets decoupled to
some extent. It also reduces transient overshoot, as enabled by bounded
inputs.
Distributed cooperative strategy for V-f regulation of lossy-line is-
Fig. 6. Design procedure of proposed control scheme (Wang et al., 2015). landed MGs is proposed in Wu, Shen, and Iravani (2017). It comple-
ments the primary droop control and achieves zero frequency error at
steady state along with optimal active power sharing. Also, maintains
In Wang et al. (2015), distributed voltage-predictive control
critical voltage level at selected nodes simultaneous to reactive power
strategy for ac MGs is proposed. It uses the estimated network voltages
sharing. The proposed scheme uses a unified dynamic small-signal MG
based on Kalman filter state estimator. The advantage of the presented
model that allows to provide control objective even when ac-side vol-
controller is being capable of providing offset-free voltage signal at a
tage limits have reached for a DER. In Lou, Gu, Xu, Cheng, and Liu
specific node. The block diagram shown in Fig. 6 gives the design
(2017) also, a secondary level V-f controller is proposed. Here, voltage
process. Several test conditions i.e. various uncertainties in parameters
control is achieved by MPC consensus tracking problem and rolling
and load, disturbances in LC filter and output impedance as well as
optimization. A PI control in addition to a global frequency estimator
faults were simulated and the result shows its robustness against all
provides frequency regulation and proportional power sharing. Here,
these issues. Later, in Wang, Wang, Chen, and Blaabjerg (2017), this
control laws are robust against intervals of information updates. Simi-
work is extended by estimating the Q distribution along with system
larly, a V-f control based on finite-time strategy is proposed in
voltages by local measurements. Thereafter, optimal controller is de-
Dehkordi, Sadati, and Hamzeh (2017). Here, advantages are that it
signed to optimize the whole MG. Although there is an improvement in
doesn't require detailed accurate model as input-output feedback line-
steady-state voltage profile, it has slower dynamic characteristics.
arization converts nonlinear, uncertain dynamics of MSs into linear. It
In Liu et al. (2016), pinning control technique is used for V-f control
is independent of parametric uncertainties of line and loads, MG con-
of MGs. It provides the advantage of using small number of feedback
figuration or feeder topology.
controller, hence by pinning certain agents of MAS, complete syn-
A droop-free frequency control scheme based on the quasi-Newton
chronization of DERs can be achieved. It is robust against commu-
technique is given in Wang, Wu, and Zhang (2017). It is a second order
nication topologies i.e. fixed and uncertain and facilitates PnP of DERs.
method hence, able to attain a super-linear convergence rate, making
Similar pinning oriented primary and secondary voltage control
faster frequency control than existing first order methods. It also
strategy with bounded uncertain time delays is proposed in Lai et al.
minimizes the generation cost by using equal increment rate criteria for
(2016). Using a DER as virtual leader, the desired frequency of rest of
power sharing among RESs. Peer-to-peer sparse communication is
DERs is pinned and based on a developed average consensus algorithm,
successfully able to tackle measurement errors, time delays and over-
Q is shared proportionally. It also improves stability, robust due to
shooting. Block diagram representation of operation of an agent is
bounded time varying communication delays. This gives an easy im-
shown in Fig. 7. In El-Taweel and Farag (2017), a novel distributed
plementation procedure with desired responses.
constraint satisfaction (DCS) based voltage control strategy is for-
Kammer and Karimi (2016) presents a primary droop and secondary
mulated. Similar peer-to-peer link and MAS is used medium to solve the
level frequency control. Here, a less complex model of multi-inverters
problem with an asynchronous weak commitment (AWC) methodology.
MG is built by dynamic phasor method, capturing the electro-
It takes into account the variations in generation and loads. This
mechanical and electromagnetic dynamics. Later, using H2 control
method is less affected by message delays and is easily scalable.
technique a secondary level scheme is formulated and solved by convex
optimization in frequency-domain. Computational burden for this
controller is relatively less, without compromising much on robust its
performance.
A peer-to-peer MAS voltage control scheme using gossip-based
communication and push-sum algorithm is proposed in
Engels, Almasalma, and Deconinck (2016). It mitigates voltage differ-
ence by updating the required P-Q at asynchronous time steps. The step
size is computed using gradient descent and backtracking line search
technique. This control strategy avoids issues of synchronization among
different devices. A secondary voltage controller that decides control
input for Q-V droop is given in Wang, Sun, and Nikovski (2016). It
provides improved power quality in MG network and maintains con-
stant voltage at PCC even during load and state variations. Similarly, a
droop based controller using two distributed impulsive algorithms is
proposed in Lu, Liu, and Zhu (2016). This gives access to set values for Fig. 7. Operational block diagram of agent (Wang et al., 2017).

126
S. Sen, V. Kumar Annual Reviews in Control 45 (2018) 118–151

However, there are other improved methods for small scale MGs and its Vi = Vi* kVi (Pi Pi*)
usage is restricted to large sized MGs only. fi = f i* k fi (Qi Qi*) (9)
A novel finite-time restoration of V-f by cooperative control based
on consensus paradigm is given in Pilloni, Pisano, and Usai (2018). It A two-layer structure for V-f control is proposed in
uses a chattering-free SMC, which enhances the system robustness and Bella, Cominesi, Sandroni, and Scattolini (2017). The primary layer
convergence properties with respect to existing designs. It achieves comprises of decentralized droop control at local DERs level to limit the
these objectives while dispensing the parameter and DERs models in- network variables within nominal range. While secondary layer uses a
formation i.e. it has a faster transient response. On a similar note, an- centralized MPC, power profiles and some predictions to formulate
other finite-time MAS based control with BESSs is proposed in Zhang constrained optimization cost function. It attains rated steady state
and Hredzak (2018). It converts the SoC balancing and voltage re- values of variables within a finite-time. However, accuracy of this
storation problems respectively to double- and single-integrator sys- control method lies in correctness of prediction technique. In Zamora
tems with some latency. Thereafter, Artstein transformation is used and Srivastava (2018), a multi-layer structure for networked MGs is
convert these to delay-free systems. It requires only sparse commu- given. Inner layers are meant to regulate V-f using LCs. The MGCC holds
nication and is resilient to link failure with PnP features. An optimal the authority to manage the EMS for power sharing with utility and
distributed V-f restoration at the secondary level is discussed in neighboring MGs. It has advantage that, if the communication link fails,
Xu, Sun, Gu, Xu, and Li (2018). Achieves finite-time convergence for droop control characteristic by lower controllers gets activated. Pro-
frequency and P sharing. The average voltage magnitude of all DERs are posed controller is suitable for distribution lines with high R/X ratio,
able to be regulated at an asymptotic convergence rate. Its benefits however, involvement of SoCs of the BESS slows down its control ac-
being cost effective and support scalability. Similarly, a secondary level tion. Recently, to tackle the issues of high cost and degradation ten-
V-f controller based on consensus algorithm is proposed in Duan et al. dencies of batteries storage units, a stochastic fractional-order PID
(2018). It computes the reference phase angles for bus voltages using based load frequency control is proposed in
fair load sharing and dc power flow. At primary level, feedback line- Khooban, Niknam, Shasadeghi, Dragicevic, and Blaabjerg (2018). It
arization with dynamic bounds on control law is designed for voltage suggests to realize the vehicle to grid (V2G) concepts using the electric
regulation and suppression of transient line current. It greatly enhances vehicles (EVs) within the microgrid. The parameters of the non-integer
system reliability under the conditions of approaching limiting capacity PID controller are tuned by the modified black hole optimization
and at some system level faults. method. Even though complexity involved in the proposed scheme is
quite high, it tracks the frequency with reduced fluctuation, deviation
and is observed to be more robust than existing load frequency con-
4.2.4. Hierarchical control technique trollers. Voltage and frequency control is summarized in Table 3.
A hierarchical structure to compensate voltage unbalances of load
buses of droop controlled MGs was first proposed in 4.3. Active and reactive power sharing
(Savaghebi, Guerrero, Jalilian, & Vasquez, 2011). The LCs of each MSs
consists primary droop control. While in secondary level, proportional- 4.3.1. Centralized control
resonant (PR) control technique is used to restore voltage amplitude at In conventional power system, task of matching generation with
load bus and frequency of whole MG to nominal value. Though this load demand is performed by economic dispatch (ED) and unit com-
controller design method is simple yet, the required level of perfor- mitment (UC). In case of MGs, generation-demand is balanced by
mance is not achieved. Three level scheme that combines low-level proper active and reactive power (P-Q) sharing algorithms. In this re-
distributed, high-layer supervision control together for economic op- gard, an interfacing converter between MG and upstream network is
eration of isolated dc MGs is proposed in (Jin, Wang, Xiao, Tang, & proposed in Deng, Zhang, Wang, Gu, and Wu (2013), Yue, Zhao, Lu,
Choo, 2014). It is accomplished by maintaining proper dc-bus voltage. and Li (2010). The flexible PE structure is formed by a back-to-back
Hierarchical Level I (HLI) is meant for keeping bus voltage within connection of full-controlled IGBT based ac-dc converters. A dc storage
nominal value through changing various modes of VSCs. The HLII controller based on PI control is placed at the middle of the converter
regulates voltage of PV modules and sets the priority of battery char- (Yue et al., 2010), while PID control in Deng et al. (2013). It helps in
ging/discharging with different SoCs. HLIII is for emergency situations maintaining accurate power flow from the grid to MG or vice versa as
like sudden load or generation change due to failure. Although it suc- per the requirement and helps to mitigate transmission of harmonic
cessfully achieves overall voltage regulation, small difference still exists components from one side to other. The advantage of these control
at bus level. A voltage control and power sharing of a cluster of dc MGs strategies for P-Q sharing is that these simultaneously accomplishes the
using a three-layer hierarchy is given in (Shafiee et al., 2014). Primary task of power exchange with upstream grid i.e. acts as partial energy
control layer locally controls the common bus voltage. While secondary management system (EMS). Also, are easy to implement.
works as: centralized to maintain nominal voltage by restoring voltage A master/slave method based two-stage centralized control strategy
deviations of every MG arising due to primary layer and decentralized for power sharing and voltage regulation of LV MG is proposed in
to maintain a power flow under inter-connected operation. The last Cavraro, Caldognetto, Carli, and Tenti (2016). Here, the master con-
layer is called power flow control (PFC) to regulate MGs’ current ex- troller located at PCC receives power data from active nodes and then
change at tie-line. Similarly, in (Lee, Zhu, Zheng, Choi, & Wei, 2014), a transmits P reference points to DERs. The data are non-critical time
controller consists of hierarchical bus, an ac-dc converter at PCC, a hence, were broadcasted over low-BW link. Slave or LC regulates the
voltage level converter and PV and BESS interfacing converter. By output voltage magnitude of DERs. Although this method achieves ac-
controlling these converters, a stable and reliable operation of all these curate P-Q sharing at PCC and also avoids overvoltage at nodes, it uses a
MGs under different modes are achieved. But, is restricted to dc type model-free method which might lead to implementation difficulties in
MGs only. actual scenarios.
A new method for V-f control is presented in Ramezani and Li
(2016). Primary droop layer is designed using direct current vector 4.3.2. Decentralized control
control (DCVC) method. Later, it integrates secondary control layer The issue of active power sharing of the VSCs in a MG is considered
with DCVC-based droop controller to achieve an improved V-f stability in Divshali, Alimardani, Hosseinian, and Abedi (2012). Decentralized
and power quality. The modified droop action is given by (9). It helps to controllers based on hybrid model is presented. It constitutes modular
prevent operation of VSIs beyond limits viz. PWM saturation limits and PI controllers for power, voltage, current and output filter control of
rated current. each DERs and works upon local measurements of frequency and

127
Table 3
Summary of voltage and frequency control techniques for microgrids.
Design technique Authors Control principle Contribution Comments (Benefits/limitations)
S. Sen, V. Kumar

Centralized control (Akbari et al., 2011) PI control A centralized voltage control method for MGs with both ac, Simple to implement, restore voltage stability under generation
dc sources under grid-tied mode. uncertainty.
(Tabrizi et al., 2012) Lead-lag controller MG voltage controller (MGVC) to handle various Improved voltage profile for both pre-and post-islanding, supports
contingencies. existing Q controllers.
(Karimi et al., 2017) Adaptive control Controlled under frequency load shedding incorporating a Modular structure facilitates easy modifications, but, use of
state estimator. estimator gives slow response.
Decentralized control (Ahn & Peng, 2013) Model-free controller Voltage control of small LV military MGs with PLL. Minimize distribution power loss, but, load buses voltages were
ignored.
(Luo et al., 2013) PID and droop control A triple-layer droop control scheme for parallel inverters. Regulates the dc-bus link, delayed control action.
(Trung, 2013) Linear Matrix Inequality (LMI) control Robust droop scheme for stable operation of MG and voltage Guaranteed zero steady-state frequency deviation, optimal power
stability. sharing.
(Liu et al., 2014) Decentralized multi-agent system Optimized average consensus Algorithm for frequency Imposed communication constraints, parameter estimation in
(DMAS) control. cooperative with controller.
(Etemadi et al., 2014) Robust servomechanism control Frequency controller by internal oscillators using GPS and a Fast tracking of set points, zero steady state error and robust
PMS for voltage and power set point. against loads variations.
(Hossain et al., 2014) Linear Quadratic Regulator (LQR) control Controller for nonlinear MG model with uncertainties and Although robust and supports islanding, but, stability region is
under islanding, fault scenarios. confined to a small post-fault operating point.
(Huang et al., 2015) PI control Two-level voltage control and parallel operation of VSCs for Enhances system reliability and avoids saturation of controller
dc MGs. efforts.
(Sadabadi et al., 2015) LMI control Voltage control of inverter-interfaced islanded MGs with Robust behaviour under load variations, accurate voltage tracking.
radial topology.
(Riverso et al., 2015) LMI control Realizing PnP capability of MSs using quasi-stationary line Includes neutral interactions for meshed MG networks without
(QSL) approximations using a global MG model.
(Toub et al., 2015) PI control with HSS tuning Voltage regulation and power flow control method for PnP as well as scalability features, improved stability.
islanded MG.

128
(Lizhen et al., 2016) Proportional resonant (PR) and PI control Local DERs controllers with two level compensation of Reduced communication needs, easy expansion, but, slower
voltage unbalances. dynamics.
(Zhang et al., 2016) Quasi-hierarchical droop and PI control Three level decentralized structure for droop controlled Simple to design, uses a nonlinear droop curve for local ED, f
islanded MGs. restoration.
(Sadabadi et al., 2017) Robust controller with convex A linear time-invariant (LTI) system with polytopic Higher failure rate as GPS is used for synchronization.
optimization uncertainty.
(Gu et al., 2017) LMI control Secondary voltage control strategy is using nonlinear state Achieves robust average voltage and facilitates accurate power
estimator. sharing.
(Cook et al., 2017) Optimal and adaptive control Voltage regulation using optimal exergy destruction (OXD) Better accuracy then droop scheme, however, design complexity is
and mode-adaptive control. quite high.
Distributed control (Wu et al., 2013) PI controller with frequency bus signaling A two layer coordinated strategy for frequency control of Simple, needs low cost, low-BW communication channels, but, has
MG. sluggish control action.
(Bidram et al., 2013; 2014) MAS with feedback linearization and Second order linear tracking problem to achieve Provides desired speed of response by parameter tuning, complex
neural network synchronized operation of multiple DERSs. design.
(Serban, 2014) PD and PID control A secondary level frequency control strategy including BESS. Simple to implement, faster transients, includes dead-zone
nonlinearity.
(Schiffer et al., 2014; Schiffer et al., Consensus algorithm Voltage control scheme based on a consensus algorithm for Applicable to arbitrary network topologies, but, assumes a small
2016) an inductive line MG. phase angle at DERs.
(Shafiee et al., 2014; Shafiee et al., Cooperative control A distributed secondary control (DSC) for regulating power, Uses wire-less discrete data packets hence, avoids data loss and
2015) voltage and frequency of a MG. time delay but, slow action.
(Chen & Feng, 2015; Guo et al., 2015) Consensus algorithm A finite-time voltage regulator and consensus-based Decoupled controllers therefore are independent, faster response
frequency control scheme. without much loss of stability
(Xin et al., 2015) Distributed incremental cost algorithm A three level structure for frequency control and ED. LPFs are designed to decouple the dynamics and enhance stability.
(Cady et al., 2015) Linear iterative algorithms and PI Frequency regulation and optimal dispatch of synchronous Event triggered, requires less control effort, zero steady state
controllers and inverter interfaced MSs. frequency error.
(Rahman et al., 2015) MAS and PI control Controller for stable load V-I during both normal operation Ease of parameter adjustments, asymptotic tracking of preset
and under faulty conditions. reference current signal.
Distributed control (Wang et al., 2015; Wang et al., 2017) Distributed voltage-predictive control Voltage regulation and reactive power sharing using state Gives offset-free voltage signal, improved steady-state but, slower
estimators. dynamic characteristics.
(continued on next page)
Annual Reviews in Control 45 (2018) 118–151
Table 3 (continued)

Design technique Authors Control principle Contribution Comments (Benefits/limitations)

(Liu et al., 2016) Pinning control and MAS Distributed architecture for V-f control of islanded MGs. Small number of feedback controller, PnP capabilities.
S. Sen, V. Kumar

(Lai et al., 2016) Pinning control and average consensus Voltage control strategy for islanded MGs with bounded Robustness to bounded time varying communication delays,
algorithm uncertain time delays. proportional Q sharing.
(Kammer & Karimi, 2016) H2 optimal control and convex A primary droop and secondary level frequency control Simpler dynamic phasor model is used, robust performance.
optimization scheme.
(Engels et al., 2016) MAS and push-sum algorithm Peer-to-peer voltage control scheme using gossip-based Asynchronous time steps and doesn't need to synchronize with all
communication. devices.
(Wang et al., 2016) Distributed droop based controller A secondary voltage controller that decides control input for Improved power quality, keeps constant voltage at PCC even
Q-V droop primary control. during load and state variations.
(Lu et al., 2016; Lu, Wang, Ai, & Lee, Distributed impulsive algorithm and LMI A secondary droop based controller for V-f restoration. Updates its states by impulse jumps in discrete time, reduce
2017) overshoots.
(Wu et al., 2017) Cooperative control A strategy for V-f regulation of lossy-line islanded MGs. Zero frequency error at steady state, uses a unified dynamic small-
signal model.
(Lou et al., 2017) MPC with consensus tracking and rolling Secondary level V-f controller with reactive power sharing. Control laws are robust against intervals of information updates,
optimization need global frequency estimator.
(Dehkordi et al., 2017) Input-output feedback linearization A V-f control based on finite-time strategy. Independent of parametric uncertainties, doesn't required a
detailed MG model.
(Wang et al., 2017) Quasi-Newton technique A second order droop-free frequency control scheme. Attains super-linear convergence rate, minimizes generation cost,
immune to measurement errors.
(El-Taweel & Farag, 2017) Distributed constraint satisfaction (DCS) A voltage control strategy with asynchronous weak Less affected by message delays, easily scalable, implementation is
and MAS commitment (AWC) methodology. restricted to large MGs only.
(Pilloni et al., 2018) Cooperative control and SMC Novel finite-time restoration of V-f of an islanded MG. Enhances system robustness, speed and convergence properties.
(Zhang & Hredzak, 2018) MAS based control Finite-time control strategy for dc MGs with BESSs. Artstein transformation converts it to a delay-free system, resilient
and PnP.
(Xu et al., 2018) Optimal control Distributed V-f restoration at secondary Level. Asymptotic convergence rate, scalable and low-BW
communication.

129
(Duan et al., 2018) Consensus algorithm and feedback A secondary level V-f controller based on consensus Accomplishes voltage regulation and suppression of transient line
linearization algorithm. current.
Hierarchical control (Savaghebi et al., 2011) Proportional-resonant (PR) control A hierarchical compensator for voltage unbalances at load Simple to realize, but, doesn't meet the required level of
buses of droop based MGs. performance.
(Jin et al., 2014) PI controller Three level strategy to combine low-level distributed and Successfully achieves overall voltage regulation, but still small
high-layer supervision for ED. differences exists at bus level.
(Shafiee et al., 2014) Consensus and PI control A voltage control and power sharing of a cluster of dc MGs. Regulate MGs current exchange at tie-line, restores voltage
deviations.
(Ramezani & Li, 2016) PI and droop control Direct current vector control (DCVC) method for V-f control Control scheme is restricted to dc type MGs only
(Bella et al., 2017) MPC A two-layer structure for V-f control of an islanded MG. Attains rated steady state values of variables within a finite-time.
(Zamora & Srivastava, 2018) Adaptive modified droop and PI A multi-layer control structure for multi-purpose networked Robust to link failures, due to SoCs of BESS slows down the control
controller MGs. action.
(Khooban et al., 2018) Fractional order PID controller Stochastic load frequency control (LFC) architecture. Reduced fluctuation in frequency and comparatively more robust
in nature but, higher complexity.
Annual Reviews in Control 45 (2018) 118–151
S. Sen, V. Kumar Annual Reviews in Control 45 (2018) 118–151

voltage. It considers one BESS unit which is being operated with a co- and last one at system level considering the generation capacities and
operative strategy to manage the dc link voltage of MSs within an ac- load types. Although this deals with actual implementation issues of
ceptable range. This control scheme is simple to realize, however, ac- hybrid MGs, several other issues i.e. interconnecting PE converters and
curacy and overall optimized operation of the MG is not achieved. To synchronization of DERs are not explored.
fill this gap, Zhao et al. (2015) presents a multi-objective control
scheme for coordination and optimization of dc MG. Here, a primary 4.3.3. Distributed control
voltage droop control is discussed first, which illustrates its load sharing Early works on P-Q sharing using the distributed scheme is explored
capabilities. Thereafter, it is augmented with a secondary integral in Bolognani and Zampieri (2011); Bolognani and Zampieri (2013). The
control to compensate for voltage deviations at steady state. A con- reactive power is compensated by using PE converters during light load
sensus filter based upon distributed averaging PI (DAPI) is designed for conditions. An approximate model of MG is used to form a reactive
regulation of optimal injections. Optimal ED algorithm for multiple power flow problem in linear constrained convex optimization frame-
DERs also achieves overall optimization. A power flow control on V-I work. Thereafter, convergence conditions and its speed, bounds are
droop characteristics is given in Golsorkhi and Lu (2014). Here, the decided by using the random gossip-like algorithm. Results show that it
voltage components of D- and q-axis is drooped as a piecewise linear is successful to minimize power distribution losses. It also has robust
function with the respective axis currents. Advantages over conven- capabilities against small load fluctuations. However, its design im-
tional P-f and Q-V droop were studied. It increases the damping while plementation is difficult and costly.
decreasing time constant hence, dynamic response of MG system and The problem of power/current sharing by the DERs and voltage
stability is significantly improved. Results show less than 3% error in regulation in a dc MG is given in Anand, Fernandes, and Guerrero
power and current sharing under high loading conditions. (2013). To accomplish the objective two control schemes are used:
A hierarchical three-layered decentralized method for active power droop to achieve proper power sharing but, it in turns deteriorates the
control using local frequency measurements, f-P droop scheme is im- voltage profile. A second controller based on digital average current
plemented in Xin et al. (2016). The controller has two parts: hier- sharing (DACS), requiring a low-BW communication is used to improve
archical frequency control (HFC) and frequency tracking control (FTC), voltage regulation. Although stable and cost effective, this controller is
realized by control PI (10). Simulations results suggest its enhanced not able to meet MGs’ operational dynamic requirements. In Asad and
efficacy along with self-optimizing capability. Kazemi (2014), a modular power sharing scheme called optimal dc
power sharing (ODCPS) for radial connected dc MGs is presented. Using
fsec ( , fi, des ) fter ( , Pi, i (Pi ))
fi, des = fpri (Pi ) + + a novel nodal pricing computation method, the detailed para-
(1 + T2 s ) (1 + T3 s ) meterization of the controller is also discussed. It also takes into ac-
with f pri (Pi ) = fmax di Pi count the power losses across the feeders to obtain an optimized control
fsec ( , fi, des ) = (fnom fi, des ) methodology. It is highly effective, and reduced computations, ro-
fter ( , Pi , bustness against load variation, expansion of DERs and least utilization
i (Pi )) = fmax di Si i (Pi ) fpri (10)
of communication.
In Yang et al. (2016), the model free current sharing scheme for dc A multi-objective two-layer control strategy is presented in
MGs with multi-buses is proposed. It is having two parts: first one is a Bidram, Davoudi, and Lewis (2014), Cai, Hu, Lewis, and Davoudi
droop controller which uses the virtual capacitor to realize a modified (2016). The first layer constitutes control scheme for voltage controlled
global indicator and rate of change of output voltage to eliminate the VSCs (VCVSCs) which main task is to regulate the frequency and vol-
effects of network mismatches. While the second one is complementary tage. While the second layer is used for current controlled VSCs
voltage stabilizer with shunt virtual resistance. These help in achieving (CCVSCs), here the objective is to achieve a proportional active and
zero steady-state voltage error. In He, Li, Liang, and Wang (2017), a P-Q reactive power sharing. This control scheme uses a one-way commu-
sharing for series-connected micro converters to integrate the DERs is nication facility and MAS method therefore, is economical than prior
proposed. The conventional droop control is not applicable for series- centralized and decentralized strategies. It is also robust against failure
connected microconverters hence a novel inverse power factor-fre- of a few communication-links. Similarly, Nasirian, Moayedi, Davoudi,
quency (PF-f) based droop controller is introduced as given in (11). This and Lewis (2015), discusses an architecture for voltage control and
scheme found to be non-sensitive to variation in feeder impedances and power sharing in dc MGs. The conventional secondary layer is replaced
filter parameters. Comparison results between the proposed and con- V-I regulators and a noise-resilient observer. It uses a global dynamic
ventional droop method show its improved performance and stability. model which communicates over sparse communication graph. This
controller has improved transient response and is quick tunable.
cut
= * + mPF . PFLPF and PFLPF = s + cut
cos( ) Thereafter, in Nasirian, Shafiee, Guerrero, Lewis, and Davoudi (2016),
with * = Ecos ( * = kinner . (I * same authors gave a droop-free multi-purpose cooperative control
dt ), V PWM Io) (11)
scheme for ac MGs. Each inverter control (an agent) has three modules:
A coordinated control of inverters operating in voltage and current voltage, reactive and active power regulator modules built-in. The
controlled modes (VCM/CCM) is discussed voltage regulator adjusts local voltage set points using estimated global
(Mousavi, Jalilian, Savaghebi, & Guerrero, 2018), for Q sharing and average voltage. Later, another reference point correction is provided
compensation of harmonic from voltage waveforms. The capacitive by Q regulator. This scheme effectively regulates P, attains proportional
virtual impedance of VCM units compensates harmonics by nullifying load sharing and synchronizes frequency by adjusting set point.
output inductance of LCL filter. The reactive power is regulated by Secondary level power sharing schemes for dc MGs are also pro-
considering distorted and remaining power capacities of both VCM and posed in Peyghami-Akhuleh, Mokhtari, Loh, and Blaabjerg (2016),
CCM units to implement modified and reverse droop strategies. Ad- Wang et al. (2016). In (Wang et al. (2016), the architecture consists of
vantages of this controller are that it gives a simple and low cost so- three controllers: voltage, current and droop coefficient controllers. It
lutions of Q-V regulation without the need for any communication uses voltage-shifting and slope-updating techniques together so that,
setup. The multi-time scale power controller for hybrid MG with several small voltage deviation is mitigated and current sharing accuracy is
interacting subgrids is proposed in Xia, Wei, Yu, Peng, and Tang (2018). enhanced during fast-changing loads. Similar droop based method is
These subgrids operate in coordination and support each other to given in Peyghami-Akhuleh et al. (2016). Here, a communication link
minimize power fluctuations. This is achieved in three time scales, first of low-BW is used to share current value information of nearby in-
two are executed at a subgrid level where the ac frequency and dc verters and adjust the output current and regulate voltage at concerned
voltage is regulated by droop control at loads and DERs respectively inverter. It consists of two local MS controllers for voltage and current

130
S. Sen, V. Kumar Annual Reviews in Control 45 (2018) 118–151

regulation. In both these methods, the average voltage is regulated by environmental conditions, fluctuating load demands. In all these cases,
using currents data. Hence, are simple to realize but, its applicability is it achieves a near-optimal solution.
limited to dc MGs only. A droop controller for sharing the harmonic and imbalance power to
A distributed control framework for VSIs in an autonomous ac MG is obtain a sinusoidal voltage is proposed in Zhou, Kim, Zhang, Sun, and
proposed in Chang and Zhang (2016). It provides a robust proportional Han (2017). Consensus control scheme is also developed to regulate the
power sharing along with frequency and voltage regulation for each virtual impedance of DERs to enhance Q sharing capability. Along with
inverter on the basis of measured P/Q power among its nearby in- these, a PI principle based secondary level voltage regulator is also
verters. It is applicable for both mesh and radial topologies with mixed developed, which compensates for the voltage deviations caused by
R/X values. Bilinear matrix inequality (BMI) constraints on voltage above two controllers. This controller has benefit of enhanced flex-
magnitudes are considered to establish convergence conditions by sol- ibility and reliability without need of information on line impedance.
ving an optimization problem at steady state. Although it has a complex Zholbaryssov and Dominguez-Garcia (2017), gives a P sharing con-
structure, benefits in terms of partial communication failure, in- troller based upon slow adjusting set point i.e. tracking a slow-varying
dependent of network topology are substantial. A secondary consensus reference so that the trajectory satisfies phase-cohesive property in
based controller for Q sharing and voltage regulation is proposed in lossless MGs. Its advantage is that the system always remain within the
Ahmadi, Nazarpour, Shafiee, and Bevrani (2016). It adds correction stable zone as line voltage phase angle always lies within |π/2|. It also
term in the droop coefficients by comparing generation of neighboring regulates frequency within nominal range. Similarly, a multi-functional
inverters, while voltage regulator maintains the average voltage around secondary control strategy is proposed in Shafiee, Nasirian, Vasquez,
rated value. PI controls are used for the purpose, whose consensus Guerrero, and Davoudi (2017). It consists of three modules: 1) voltage
based tuning and parameters of communication link is set using fuzzy regulator, which regulates the average MG voltage at nominal value. 2)
logic. This method helps to include parameter uncertainties within reactive power regulator, diminishes Q mismatch by tuning the coef-
various elements of the MG. In Hu, Zhou, Liu, Guan, and Chi (2016), ficients of Q-V droop. 3) P-f regulator, maintains nominal frequencies of
work from Bolognani and Zampieri (2013) is extended further by in- all DERs along with P sharing. It doesn't need f measurements as it uses
cluding generation limits of MSs into reformulated objective function. active power data to control frequency. If any communication link fails,
Then, by using logarithmic barrier function, optimal control framework it is able to adapts and reduces own capability to acts as a conventional
is converted into equality constraint problem. This not only eases droop controller.
computations for control action, resulting into faster response but also, Distributed estimation and secondary controller (DESC) for islanded
enhances its accuracy. MGs is proposed in Zhang et al. (2017). It achieves active power sharing
A reactive power sharing technique with unmatched output im- and frequency normalization by using two subsystems: (1) dynamic
pedance of DERs is given in Zhang, Kim, Sun, and Zhou (2017). It es- average observer (DAO) estimates the average load demand, which is
timates reactive power mismatch and uses a PI distributed consensus compared to total generation of PVs to compute the power unbalances.
scheme to adaptively adjust virtual impedance of the DERs. By doing so Thereafter, using (2) finite time consensus controller, active power is
the circulating line current is also getting suppressed. Another con- shared fairly within a small time scale. Its accuracy is more and ac-
sensus controller restores average voltage of MG by regulating output complishes frequency regulation but, inclusion of an estimator also
voltage of each DER. This gives an extra DOF for developing voltage creates stability issues and increased complexity. A fully distributed
controller along with proportionate Q sharing. Similar decoupled ar- strategy is discussed in Meng, Wang, and Liu (2018) for intermittent
chitecture is discussed in Fan, Hu, and Egerstedt (2017), for islanded RESs using a need-based aperiodic data transfer. It maintains supply-
MGs where the voltage controller works with a nonlinear state feed- demand by MAS-based discovery algorithm to acquire the knowledge of
back, built using the Lyapunov principle. Here, MGs’ inverter model for whole MG. The benefit of this scheme is that as the agents transfers its
injection of reactive power is built based on bus angle differences. This status to neighbours only when an event occurs, so reduced capacity
reduces the amount of communication network usage by using event- communication links are enough. Similarly, in Cai and Hu (2018)
triggered control scheme. Also, because the droop control is considered power control within the ac MG with cluster of spatially concentrated
hence, avoids secondary control layer, therefore, is faster and more dispatchable DERs is discussed. The problem is formulated as a leader-
accurate. Similarly, Moayedi and Davoudi (2017), presents a controller follower MAS based tracking at two levels. The lower controller is ro-
to achieve both proportional power sharing and bus voltage regulation bust in achieving set power output tracking by taking account of un-
simultaneously. The total generation cost is minimized by solving an ED certainties and upper one decides the proportional power sharing base
problem to accomplish generation optimization. Later, voltage re- for these DERs using global information. It takes into account uncertain
ference of VSCs are adjusted by (1) voltage regulator to match gen- dynamics of DERs using internal model design and supports PnP cap-
eration-demand equality (2) optimizer to have same incremental costs. ability.
Then, it is used to enforce inequality constraints. Overall design is
difficult and has stochastic interferences. But, it accomplishes multiple 4.3.4. Hierarchical control
tasks. The objective function is given as: In a first, two-level MAS based hierarchical control strategy for
regulation of P-Q under both operating modes is presented in Chamorro
N N
and Díaz (2013). The lower physical layer is for the droop control of
L= Ci (Pi ) + PD + PL PND Pi
i=1 i=1 (12) VSCs depending upon load changes and power exchanges. Independent
nonlinear fuzzy controllers and two PD fuzzy controllers are used at this
In (Wang et al. (2017), issues of reactive power and voltage reg- level. The upper layer sends the set points values for the lower layer
ulation is tackled by a combination of a Kalman based state estimator agents based upon dynamic observation and pre-specified rules. Im-
and distributed optimal controller. An input-output relationship for the portant features like reconfigurable, fast synchronization, are added.
whole MG system is built using small-signal model and sensitivity However, the conversion of the nonlinear fuzzy controller to linear PD
analysis to evaluate the voltage and power sharing response. Results is a tedious task. A two-stage frequency management system of an is-
suggest that, its dynamic response depends on the coefficients of power landed MG is proposed in Rezaei and Kalantar (2014). The first layer
error and voltage weight. In Li et al. (2017), a two-layered network is utilizes stochastic random scenarios depending upon Roulette Wheel
presented. The communication network has two sub-graphs: first one Method (RWM) and Lattice Monte Carlo Simulation (LMCS) including
includes all agents, which maintains the generation-demand balance intermittent sources and load fluctuation. Later, second stage has MILP
and later one constitutes of only controllable agents that ensure optimal objective function by taking into account the economic policies, various
dispatch of P-Q. This scheme is evaluated by different cases of stochastic scenarios, contingencies of unit or line outage etc. however,

131
S. Sen, V. Kumar Annual Reviews in Control 45 (2018) 118–151

its boundedness within predefined ranges is not quite effective. sharing are managed. However, the dynamic response is observed to be
In Caldognetto, Tenti, Mattavelli, Buso, and Brandao (2015), a sluggish in nature. Similarly, two-level optimal power sharing, voltage
three-layered master/slave control architecture is given. It uses con- regulation, optimization of transmission loss in dc MGs is proposed in
servative power theory (CPT) at primary level to manage local loads by Ma, Yuan, Zhao, and He (2017). The primary layer consists of a voltage
compensating its reactive and harmonic currents to stabilize the vol- droop controller for each VSCs. The secondary layer aims to realize
tage. P-Q sharing is managed at secondary level using MCs acting as optimal power flow to the loads. Advantage of this is that it doesn't
slaves, located with DERs. Interactions between MG and utility is need prior information about grid's conductance and load distribution
managed at tertiary level using a master controller. Usage of CPT helps matrices. But, its application is restricted to dc MGs only. Table 4 gives
to bring a certain level of optimization. A two-level load sharing, power complete summary of active and reactive power sharing for MGs.
quality improvement scheme is proposed in Golsorkhi, Lu, Savaghebi,
Vasquez, and Guerrero (2016). Here, the primary layer comprises 4.4. Energy management system (EMS), market participation
adaptive V-I droop controllers, where coefficients are adjusted as a
function of peak output current of DERs. A GPS timing based frequency 4.4.1. Centralized control
synchronizer is also used. Secondary layer has a mechanism to elim- Main objectives of energy management system (EMS) is that it
inate voltage harmonics at sensitive buses by generating a compen- maintains smooth operation within a MG by exchange of surplus or
sating signal, hence, power quality is enhanced. Issues of using GPS is deficit power from upstream network, along with creating earning for
equally applicable here. the MGs owner. As generally a microgrid is meant to operate under
A novel ANN based reactive power set point calculation for im- grid-connected mode, it is essential to study the market participation of
proved power sharing in conventional hierarchical controlled MGs is the MG. Several MGs within small localities can be combined to form a
presented in Baghaee, Mirsalim, and Gharehpetian (2016). The Radial multi-MGs system thereafter seek permission from the distribution
Basis Function (RBF) is applied on classical power flow equations, system operator (DSO) or contact a registered aggregator to sell power
which is faster and accurate as well as able to enhance stability margins into the LV distribution network Parhizi, Khodaei, &
under both small and large signal disturbances using an additional Shahidehpour, 2018). Hence, various concept of electricity market and
complementary control loop. However, usage of ANN increases overall demand response management has to be studied in relation to MGs. As
complexity and implementation difficulties for the controller. An en- this article is meant for control of MGs further discussion on these are
hanced P-Q sharing for islanded ac MG is proposed in Han et al. (2016). out of the scope and interested readers can find more on role and
The primary layer implements three loops viz. virtual positive sequence working of aggregators, DSO in presence of MGs in Meng et al. (2016),
impedance (VPI) loop that helps to achieve P-f droop and virtual ne- Nosratabadi, Hooshmand, and Gholipour (2017),
gative sequence impedance (VNI) loop for Q-V droop at fundamental Palizban, Kauhaniemi, and Guerrero (2014) and the references therein.
frequency. The virtual variable harmonic impedance (VVHI) enables The centralized control i.e. use of MGCC is done extensively to carry out
harmonic power sharing features at specific harmonic frequencies. this objective. Early works on use of MGCC for load curtailment during
These frequencies are extracted by delayed-signal cancellation using a peak hours for an islanded MG is given in Zaidi and Kupzog (2008). At
second order generalized integrator (DSC-SOGI). This achieves high the local level, it introduces control nodes (comprising power mea-
accuracy of P-Q sharing, power quality with cost effective low-BW links. suring unit, wireless sensor and switch) connected with loads that to be
A closed loop multi-layer control strategy for optimal power sche- converted into intelligent loads and MGCC uses them as per priority and
duling and voltage regulation is proposed in control decision. This is also known as automated load management
Deckmyn, Vandoorn, Vyver, Desmet, and Vandevelde (2017). The and easy to implementation. In Tsikalakis and Hatziargyriou (2008),
higher one called scheduling layer performs the day-ahead planning of economic optimization and market participation of MG with upstream
MG to minimize operating cost by UC for next 24 h. In middle execution grid are studied in detail. Optimization problem is formulated for P
layer, the off line power flow analysis is done to get initial bus voltage only, Q and security constraints were ignored. The demand side bidding
values (see Fig. 8). Lastly, lower adjustment layer maintains voltage for non-critical loads (shiftable and interruptible) were included in two
profile of MG by providing dynamic gains for DERs. This layer provides market policies: operational cost minimization and maximizing profit
feedback to the upper layer and iteratively voltage deviation and power (see Fig. 9), as represented in ((13).
N L
Cost = i=1
active _bid (x i ) + AX + j=1
load _bid (yj )
N N N
Profit = A x
i=1 i i=1
active _bid (x i ) i=1
load _bid (yj ) (13)
An optimization and dispatch algorithm for MG is given in
Ding, Zhang, Mao, Yang, and Liu (2010). Four objective functions were
presented which minimized operational as well as depreciation cost,
while maximizing the comprehensive and environmental benefits.
Here, overall system is optimized, but, individual operation are only
able to achieve a sub-optimal solution. Also, handling of different ob-
jective functions increases computational burden and slows down the
process. An EMS based on the centralized control is built for an islanded
MG in Olivares, Cañizares, and Kazerani (2011). Whole problem of
energy management meant for complex mixed-integer nonlinear pro-
gramming (MINLP) is divided into two simpler sub-problems of UC by
mixed integer linear programming (MILP) and optimal power flow
(OPF) using nonlinear programming (NLP). This controller simplifies
the complex process without compromising much on accuracy. Simi-
larly, an optimal EMS for an isolated MG is discussed in
Olivares, Cañizares, and Kazerani (2014). Here, has two parts: UC and a
multi-stage Economic Load Dispatch (ELD) (see Fig. 10). It takes into
account the generation and load forecasting data and SoCs of BESS.
Fig. 8. Proposed multilayer control structure (Deckmyn et al., 2017). Later, calibration, tuning, impacts of forecasting and performance

132
Table 4
Summary of active and reactive power sharing techniques for microgrids.
Design technique Authors Control principle Contribution Comments (Benefits/limitations)
S. Sen, V. Kumar

Centralized control (Yue et al., 2010) Proportional-Integral (PI) control An interfacing converter between the MG and upstream Maintains accurate power exchange between grid and MG, easy to
network. implement.
(Deng et al., 2013) Proportional-Integral-Derivative (PID) Control of interfacing converter for grid-connected MGs. Mitigates transmission of harmonic components from one side to
control other, P-Q sharing.
(Cavraro et al., 2016) Model-free master/slave control A two-stage strategy for power sharing and voltage regulation Achieves accurate P-Q sharing, avoid overvoltage at nodes.
of LV MGs.
Decentralized control (Divshali et al., 2012) PI control Modular controllers for power, voltage, current and output Manages dc-link voltage using one BESS, simple but, not accurate and
filter of each DERs to have P sharing. nor optimized.
(Golsorkhi & Lu, 2014) V-I droop control A new droop based power flow control of islanded MGs. Increases damping while decreasing time constant, better accuracy.
(Zhao et al., 2015) Distributed averaging PI (DAPI) A multi-objective control schemes for coordination, Optimal ED is achieved, minimum voltage deviations at steady-state.
control optimization of dc MG.
(Xin et al., 2016) PI control f-P droop scheme to achieve active power control of DERs Better efficacy and self-optimizing capability.
using local frequency measurements.
(Yang et al., 2016) Model-free control Decentralized current sharing scheme for dc MGs with multi- Achieves zero steady-state voltage error, simple to realize.
buses.
(He et al., 2016) PF-f droop control A P-Q sharing for series-connected micro converters to Non sensitive to variation in feeder impedances and filter parameters.
integrate DERs.
(Mousavi et al., 2018) Coordination control Regulation of VCM/CCM inverters for reactive power sharing Simple and low cost solutions of Q-V regulation, adaptive virtual
and harmonic compensation. impedance as considers set point changes.
(Xia et al., 2018) Coordination and droop control A multi-time scale power controller for hybrid MG with Includes generation capacities and load types, but, interconnecting
several interacting subgrids converters synchronization of DERs are not done.
Distributed control (Bolognani & Zampieri, 2011; 2013) Convex optimization and random First work on P-Q sharing and Q compensation using the Minimize power distribution losses, robust to load fluctuations but,
gossip-like algorithm distributed scheme costly and difficult to implement.
(Anand et al., 2013) Droop and digital average current Power/current sharing by DERs and voltage regulation in a dc Although stable and cost effective, doesn't meet dynamic
sharing (DACS) MG. requirements.

133
(Asad & Kazemi, 2014) Modular optimal pricing control A modular technique called optimal dc power sharing Reduced computations, robust against load variation, easy expansion
(ODCPS) for radial connected dc MGs. of DERs.
(Bidram et al., 2014; Cai et al., 2016) MAS A multi-objective voltage and current controlled strategies for Uses one-way communication facility, hence, economical and robust
inverter operated MGs. to communication failures.
(Nasirian et al., 2015; Nasirian et al., Droop-free cooperative control Voltage control and power sharing in dc and ac MGs Effectively regulates P, attains proportional load sharing and
2016) respectively. synchronizes the frequency.
(Wang et al., 2016) Averaging and PI control A secondary level power sharing schemes for dc MGs. Small voltage deviations are mitigated and current sharing accuracy is
enhanced.
(Peyghami-Akhuleh et al., 2016) PI controller A droop based method with two local MS controllers for Average voltage is regulated by using currents data, simple but, usage
voltage and current regulation. is only for dc MGs.
(Chang & Zhang, 2016) Bilinear matrix inequality (BMI) A distributed control framework for robust power sharing of Applicable for both mesh and radial topologies with mixed R/X
control VSIs in an autonomous ac MG. values, complex structure.
(Ahmadi et al., 2016) Consensus based tuned PI controllers A secondary level consensus based controller for Q sharing Take account of parameter uncertainties, communication channel is
and voltage regulation for islanded ac MGs. set by fuzzy logic.
(Hu et al., 2016) Optimal control Improved P-Q sharing including the generation limits of MSs. Eases computational burden, results into faster response and
enhanced accuracy.
(Zhang et al., 2017) PI based distributed consensus control A reactive power sharing technique under unmatched output Suppresses circulating current, adaptive virtual impedance, additional
impedance of DER interfacing inverters. DOF.
(Fan et al., 2017) Nonlinear state feedback controller Decoupled architecture for voltage control and Q sharing. Event-triggered control, reduces communication usage, more
accurate.
(Moayedi & Davoudi, 2017) Optimal control A scheme to simultaneously achieve proportional power Optimizer enforces inequality constraints, complexity is more due to
sharing and voltage regulation for dc MG. stochastic interferences.
(Wang et al., 2017) Optimal control Reactive power and voltage regulation by a Kalman based Dynamic response depends on coefficients of power error and voltage
state estimator and distributed controller. weight, slow action.
Distributed control (Li et al., 2017) MAS A two-layered network meant for optimal dispatch of power Two-graph communication network is used, achieves a near-optimal
in a MG. solution.
(Zhou et al., 2017) Consensus and PI control A droop controller for sharing the harmonic and imbalance Enhanced flexibility, reliable as doesn't need information of line
power to obtain a sinusoidal voltage. impedance.
(continued on next page)
Annual Reviews in Control 45 (2018) 118–151
Table 4 (continued)

Design technique Authors Control principle Contribution Comments (Benefits/limitations)

(Zholbaryssov & Dominguez- Distributed networked algorithm P sharing controller using slow adjusting set point satisfying Always remain within stable zone, simultaneously regulates
S. Sen, V. Kumar

Garcia, 2017) phase-cohesive property in lossless MGs. frequency.


(Shafiee et al., 2017) PI controller A multi-functional secondary control strategy for ac MGs. Doesn't need f measurements, works as droop controller if link fails.
(Zhang et al., 2017) Consensus based control Distributed estimation and secondary controller (DESC) for Accurate, accomplishes f regulation, but, has stability issues, more
islanded MGs. complexity.
(Meng et al., 2018) MAS-based discovery algorithm A fully distributed power sharing strategy for intermittent Need-based aperiodic (event based) data communication, includes
RESs. uncertainties.
(Cai & Hu, 2018) Leader-follower MAS and robust Power control within the ac MG with cluster of spatially Robust in achieving reference power output tracking, uses internal
control concentrated dispatchable DERs. model design for simplification.
Hierarchical control (Chamorro & Díaz, 2013) MAS with fuzzy tuned PD controller Two-level strategy for regulation of P-Q under both operating Reconfigurable, fast synchronization, however, controller design is
modes. complex.
(Rezaei & Kalantar, 2014) Mixed-integer linear programming Two-stage frequency management system and P-Q sharing of Considers various stochastic scenarios, contingencies, but, bounded
(MILP) an islanded MG for predefined range.
(Caldognetto et al., 2015) Adapted conservative power A three-layered architecture using conservative power theory Acts as EMS also at master level, bring a certain level of optimization.
algorithm (CPT) for compensation of Q and voltage.
(Golsorkhi et al., 2016) Adaptive V-I droop controllers A two-level structure for load sharing and power quality Eliminates voltage harmonics, failure issues arises due to use of GPS.
improvement in isolated ac MGs.
(Baghaee et al., 2016) Artificial neural network (ANN) based Novel ANN based Q set point calculation to improve power Enhances stability margins, is complex and has implementation
controller sharing. difficulties.
(Han et al., 2016) Proportional plus multi-resonant An enhanced P-Q sharing in an islanded ac MG. Highly accurate as suppresses harmonics, uses cost effective low-BW
controller links.
(Deckmyn et al., 2017) Optimal control and GA A closed loop multi-layer strategy for optimal power Serves day-ahead planning of MG, minimizes operating cost by unit
scheduling and voltage regulation. commitment.
(Ma et al., 2017) Optimal control Two-level optimal power sharing and optimization of Doesn't need prior information about grid's conductance and load
transmission loss in dc MGs. distribution matrices.

134
Annual Reviews in Control 45 (2018) 118–151
S. Sen, V. Kumar Annual Reviews in Control 45 (2018) 118–151

Papadaki (2015). Five agents were designed for each component of MG


i.e. RES, Battery, desalination, electrolyzer and fuel cell, so that mon-
itoring and control achieved and the MG is optimized. The theory of
fuzzy cognitive maps (FCM) is used for decision making. Particle swarm
optimization (PSO) is used to optimize working of agents for co-
ordinated operation. This decentralized EMS (DEMS) is compared with
centralized EMS (CEMS). Results suggesting performance of DEMS is on
par CEMS and in long run DEMS is economical. Similarly, in
Chen, Chen, Guerrero, and Abbott (2016), two-layer MAS based con-
troller is proposed. Lower layer constitutes the complete electrical
distribution of MG and its upper layer comprises of agents based net-
worked communication structure. These agents decide the control laws
for adjusting the DERs output as per local information. Though the
Fig. 9. Information flow between MGCC and local controllers. decoupled nature is easy to operate, it lacks the coordination to facil-
itate market participation of MG. In Xia, Peng, Hu, Wang, and Wei
(2016), a unified controller for dc MGs is presented. It serves various
purposes i.e. in grid-tied mode it can act as an EMS and provide ED by
injecting specified output current and during islanded mode, accurate
load sharing and directly regulates average bus voltage without any
extra secondary control loop. PI scheme is used for inner and outer
current control loops. This successfully presents a low cost control
scheme whose unified nature attains several attributes of centralized
EMS.
A demand response management algorithm is presented
inSakurama and Miura (2017). It uses a novel smart meters commu-
nication network topology based decentralized pricing. It computes
Fig. 10. Block diagram of centralized EMS.
proper control signals by analyzing the prices and incentives. Optimi-
zation based on Lagrangian algorithm is implemented. However, its
evaluation are discussed. Results indicating a certain delayed dynamic overall structure is highly complex. In Nutkani, Loh, Wang, and
response, which is due to the inclusion of SoCs. Thereafter, Ambia, Al- Blaabjerg (2017), a comprehensive ED method is proposed. Here, the
Durra, Caruana, and Muyeen (2014) presents another central super- dispatch priority and droop coefficients of DERs are decided by taking
visory Control (CSC) unit meant for power management of hybrid MGs into account various factors viz. voltage, power and frequency limits of
comprising PVs, wind turbines and BESS. Here, the transient response MSs, generation cost and a number of participating DERs in economic
obtained is faster as multiple PE converters are used both in grid and operation, and the required minimum online power reserves. Experi-
locals for smoothening out power fluctuations. ments with three dispatchable MSs shows enhanced cost efficiency. It
In Hajar, Hably, Bacha, Elrafhi, and Obeid (2016), a central MPC also facilitates step-load changes within MG as desired online power
based control system called global management controller (GMC) is reserves can be set conveniently.
given. It works similar to EMS for a distribution network that co- A load sharing and power management scheme with hybrid PV/
ordinates and optimizes the operation of interconnected MGs as well as battery units for MGs is proposed in Karimi et al. (2017). By considering
of distributed system operator (DSO). It includes forecasting data on the amount of available power from PV, load requirements, battery
power generation, load demand and energy prices. The functioning of conditions i.e. SoC, and the three modes of MG operations, the working
RESs and BESSs are also handled using control unit of GMC. It has the of each unit is divided into five states depending upon mismatches in
benefit of covering a larger geographical area. In Hajar et al. (2016), demand and generation. Modified P-f droop is used for load sharing
same authors proposed nonlinear optimization algorithm to formulate among the units according to operating states subject to the constraints
EMS with MPC. Inclusion of forecasts information improves the re- of power ratings. (14) presents functioning of the proposed method by
presentation of actual operating scenario, but, the overall design is PI control. The switching between states is achieved via frequency level
highly complex. A heuristic master-slave architecture under both op- triggering hence, requires less control efforts.
erating modes is discussed in Almada, Leao, Ampaio, and Barroso
(2016). During islanded operation, the PEI for BESSs acts as a master
converter which defines V-f reference set points for remaining PEIs.
(
f = f0 + mp +
mi
s ) (P ref Pout )

Similarly, during grid-tied operation PEIs at RESs plays masters’ role,


provides V-f set points at PCC. It also includes various uncertainties of
(
Pref = mp +
mi
s ) (V dc V *dc)
(14)
RESs and varying electricity tariffs along the day. Hence, opens new
Further details and definitions are available in Karimi et al. (2017).
scopes for active market participation of RESs based MGs.
The authors later extents this decentralized technique to hybrid MGs for
both single phase and three phase connections
4.4.2. Decentralized control (Karimi, Oraee, Golsorkhi, & Guerrero, 2017). A multi-functional robust
A decentralized EMS for coordinated operation of RESs and BESSs strategy for hybrid ac/dc MGs with nonlinear and unbalanced loads is
using smooth switching droop control (SSDC) is given in proposed in Baghaee, Mirsalim, Gharehpetian, and Talebi (2017). A
Wu, Tang, Dragicevic, Vasquez, and Guerrero (2015). It uses four dif- SMC controller for dc/dc converters for achieving proportional P-Q
ferent operating modes where RESs and BESSs work either in voltage sharing along with harmonics control presented. Thereafter, a two-level
control mode (VCM) and power control mode (PCM) based on classical hierarchical positive sequence power control using SMC and a negative
PI/PID control principles. SSDC scheme uses frequency bus-signaling sequence current control using the Lyapunov function theory for ac side
(FBS) to decide, trigger mode switching. It achieves advantages of both of DERs is presented. These control strategies are independent of MG
bus-signaling and droop control methods. However, it fails to economic topology and load characteristics.
optimization of overall MG. A multi-agent based EMS for MG with
various RESs is presented in Karavas, Kyriakarakos, Arvanitis, and

135
S. Sen, V. Kumar Annual Reviews in Control 45 (2018) 118–151

4.4.3. Distributed control


A control technique for PnP of RESs and reduction in power dis-
tribution loss is proposed in Costabeber, Tenti, Erseghe, Tomasin, and
Mattavelli (2011). Here, the communication is based on power line
carrier (PLC). Each active node transfers token (information packets)
sequentially to other active nodes, which is known as the token ring
method. Principle of optimal control and dynamic grid mapping fea-
tures were used, where each active node along with its neighboring
nodes are identified in real time and acted upon. Here, although cheap
means like PLC is used, additional infrastructure or setup is needed to
developed, which is not cost effective. The power management between
two interlinked MGs by optimal number of interlinking inverters is
proposed in Nutkani, Loh, and Blaabjerg (2012). The Q-V and P-f droop
schemes are used for controlling reactive and active power in the MGs
individually. For peak hours, interlinking inverters are optimally con-
trolled for reactive power support, while during light load conditions
active power is transferred among MGs. These inverters are turned off,
to minimize power loss if there is no usage. In Xu, Zou, and Niu (2013),
a distributed MPC as an optimal EMS that achieves a minimum excess/
insufficient energy in each MG for an interconnected MGs system is
proposed. These multi-MGs exchange power, data to maximize overall
energy efficiency. Required constraints are also included to achieve
guaranteed closed-loop stability. The prediction model uses one-step
delay, makes it able to capture more practical scenario. In Shafiee et al. Fig. 11. Flowchart of optimization method (Liu et al., 2015).
(2014), a consensus algorithm based on dynamic averaging for power
flow control among cluster of dc MGs is proposed. The reference is set optimal power injections. Another advantage is that it also achieves ED
by a cooperative algorithm that takes into account SoC of BESSs inside by formulating an optimization problem based on quadratic with linear
each of the MGs. It also presents a voltage controller to mitigate steady- constraints. Similarly, a load shedding problem for islanded MGs based
state voltage deviations in each MG. Also, an adaptive droop control is on distributed cost optimization (DCO) technique is proposed in
implemented, where coefficients are chosen to equalize SoC. This Liu et al. (2015). It consists of two layers: first being an improved
achieves extended feature of optimal storage, however, proposed average consensus algorithm (IACA) for global information sharing of
scheme doesn't facilitate grid interaction. monitored total active power deficiency and second being DCO using
On a similar note, Hossain et al. (2014), Hossain, Mahmud, Milano, MAS for load sharing by including overall and marginal costs. The
Bacha, and Hably (2016) presents a distributed controller that mini- overall optimization flowchart is shown in Fig. 11. It not only provides
mizes the negative or unnecessary dynamic interactions that occurs an accurate optimal load shedding but also meets PnP needs.
among multiple islanded MGs, while controlling optimal power flow. A cooperative multi-purpose control scheme for V-f control, power
BESSs and STATCOMs are used to regulate P-Q power, respectively. sharing, MG and utility grid synchronization is proposed in
Proposed control is based on H2 and LMI strategy. Complicated steps Hamidia, Livania, Hosseinianb, and Gharehpetian (2015). At primary
are required to be addressed to implement the controller however, layer droop characteristics is used and its set points are decided by
optimal response is resulted. Combined optimization of interconnected- cooperative controllers at secondary level, to minimize V-f deviation
MGs (IMGs) to minimize the operating cost of multi-MGs is proposed in and proportional power sharing among the MSs at tertiary level. Al-
Li, Liu, and Zhang (2015). It takes into account ultra-short term fore- though this controller is successful in performing aforementioned
casting and further provides active power reference point for the dy- functions within the MG, its working as EMS is not explored beyond just
namic droop controller, which works based on principle of equal gen- reconnection and synchronization. In Stadler, Ashouri, and Marechal
eration incremental cost and PI methods. Design is simple and (2016), MPC based control strategy for coordinated operation of DERs
forecasting improves practical MG scenario. A control of P sharing is presented. The local MPC algorithms acts independently to find op-
between MG and utility grid is presented in Wang, Wu, and Zhang timal reference points of controllable DERs, which in turn allows global
(2015). It uses principle of equal incremental rate which attains the independent system operator (ISO) work as virtual power plant (VPP).
minimum generation cost. Speed of controller is enhanced by updating It improves ancillary services provided to DSO. As MGs’ demand fore-
the control signal with latest local measurements using iterative and casting is included, provides day-ahead load serving and peak saving.
distributed sub-gradient method. Communication links are of low-BW A two level ED problem along with voltage regulation for a dc MG is
as only incremental cost is exchanged by each agent. In (Zaery et al., proposed in Zaery, Ahmed, and Orabi (2016). At the tertiary level,
2015), a cooperative controller that minimizes the power generation normalized operating cost of DERs is put through dynamic consensus
cost in a dc MG is discussed. It compares the instantaneous generation based algorithm to solve ED for optimal load sharing among DERs. At
costs of each MS with weighted average of neighboring DERs. There- secondary level, average voltage of MG is controlled to satisfy the
after, adjusts the voltage reference point of local MS by generating a constraint of supply-demand balance. This scheme ensures lower total
voltage correction term as per the result. Benefit of this scheme is that, generation cost (TGC) and enhanced stability. It also provides PnP
it leads to production of more power by cheaper MSs. Also, as the ro- features. Economic operation of a power park having several MGs is
bust spanning tree communication graph is used, prior knowledge of proposed in Xiao, Zhang, Wang, and Xu (2016). The modified droop
number of MSs present in MG is not required. scheme based upon the frequency/incremental cost (f/IC) of each MG is
A multi-objective control strategy for dc MGs is proposed in Zhaoa used, which is also validated in comparison to conventional f/P droop.
and Dorfler (2015). It presents a primary droop controller that utilizes a Results suggest its ability to successfully ensure system reliability, is
linear model, to achieve proper load sharing. However, it gives rise to more economical and has a faster response. A novel EMS based upon
voltage drifts at steady-state, which is compensated by two control dynamic population game theory is given in Mojica-Nava, Rivera, and
strategies at the secondary level. A PI controller and consensus filter on Quijano (2016). It uses distributed replicator dynamic method for ED
communication link performs the task of mitigating voltage drift and

136
S. Sen, V. Kumar Annual Reviews in Control 45 (2018) 118–151

considering power losses in feeders, over specific communication to- complete design process is based upon the small-signal stability study of
pology. It employs two heuristic optimization algorithms to estimate developed MG model, this gives improved stability of complete system.
network loss coefficients under various demand patterns. Although it is In Bersani, Dagdougui, Ouammi, and Sacile (2017), the problem of
able to regulate frequency and simultaneously facilitate ED scheduling, power flow among IMGs is tackled using robust controller based on
but, implementation difficulties in real scenario related to heuristic cooperative decision principle. It helps to enhance service by mini-
search pattern is quite high. mizing interaction with main utility grid by controlling maximum de-
A control strategy that combines the secondary level, small time- viation from an agreed power exchange along with pre-set values of
scale automatic generation control (AGC) and large time-scaled tertiary storage units. The problem formulation is done using partially nested
level ED is given in Li et al. (2016). Proposed scheme is called dis- information. In Li et al. (2018), a consensus averaging based algorithm
tributed economic AGC (EAGC) algorithm, and is based upon optimal for control and management of islanded MG is discussed. Here, dis-
control method. It is capable to work in all modes of operation. Com- tributed and networked MAS is used to design the control law, which
munication topology is designed by satisfying the N-1 rule, which in- has two terms: dynamic and fixed weights. The term having dynamic
creases system reliability. Similarly, a secondary control for optimal weight adjusts the output of DERs to reach different desired targets. The
power sharing and transmission loss reduction for dc MG is proposed in main advantage of this method is that during iterations the system
Chen et al. (2016). It consists of three parts: (1) consensus based voltage balance (obtained by primary layer) is never broken. A controller for
regulation, compensating primary droop caused voltage drift by gen- hybrid MGs to attain a global economic operation is proposed in
erating voltage offset. (2) proportional current control, which attains Lin et al. (2018). It consists of two levels: lower one has ac frequency-
accurate power sharing during heavy load conditions. (3) tie line cur- incremental cost (f-IC) droop, dc bus voltage-incremental cost (V-IC)
rent control, used under light load conditions to avoid transmission droop realized at ac and dc subgrids, respectively. Later, a distributed
losses. Although it is only limited to dc MGs, the analytical solutions control canonical form (DCCF) is used to recover deviations at upper
found on power flow with impacts of line resistance and drifts are level. It also suggests a novel relative loading index (RLI) to extract any
highly efficient. hidden loading conditions which help to easily define the references for
In Xu et al. (2016), an ED problem of multi-MGs with PV-storage interlinking PE converters.
combination is proposed. The distributed algorithm includes MPC
strategy along with time-sharing price to fix time and rate of power
4.4.4. Hierarchical control
supply to meet instantaneous load demand. During grid-connected
For EMS, an intermediate controller placed in-between the DSM and
operation, economic results are obtained by solving optimal objective
MGCC hierarchically to have an optimal frequency regulation in multi-
function using generation-demand predictions. While in autonomous
MGs system is presented in Gil and Lopes (2007). It is called central
mode the multi-objective function is solved using MGCC of each MGs
autonomous management controller (CAMC) which is under the au-
and receding-horizon optimization technique. Overall controller
thority of DSO. It determines the change in generation set points of
strategy is simple and satisfactorily achieves desired objectives opti-
DERs at MGCC based on frequency variation within MGs by using PI
mally. A control strategy to combine functions of secondary level
control principle. Thereafter, in MGCC power generation by MSs is
maintain voltage profile and tertiary level minimized generation cost,
optimized based upon a standard linear optimization method. Though it
enhancing utilization of renewable sources in a dc MG is proposed in
is simple to realize, dynamic response is observed to be sluggish.
Wang, Wu, and Zhang (2016). To achieve these, it uses a sub-gradient
On a similar note, in Ding, Ma, Bi, Mao, and Chang (2013), a three
algorithm and equal incremental rate (ICR) criteria. Here, only sparse
layered control architecture for energy coordination of isolated multi-
communication network is required to exchange generation ICRs be-
MGs is proposed. The lower layer is called distributed generation level,
tween neighbours. This design has additional features like scalable and
having three agents: generation, storage and load agents, which con-
flexible, reliable to communication failures and delays. In
trols respective devices at LV level. The intermediate layer is known as
Dorfler, Simpson-Porco, and Bullo (2016), distributed architecture
MG level, with four agents: MGCC, generation, storage and load agents
based upon averaging PI principle for ED at tertiary level is presented. It
working at MV level. The upper layer called multi-MGs level has a
establishes the need for droop control to attain economic optimization
CAMC agent. All these agents were based on the foundation for in-
by selecting droop coefficients for obtaining proportional power
telligent physical agents (FIPA) which improves the speed of operation.
sharing among the DERs. Here, DOF for EMS is more and discrepancies
An operation and control of MGs with frequency stability is given in
are well compensated. A novel EMS for multi-MGs based upon dis-
Nakazawa, Murakami, Shimasaki, Hayashi, and Nakanishi (2013). The
tributed MPC (DMPC) and economic MPC (EMPC) are given in
RESs are considered as disturbances and using short term forecasting
Kou, Liang, and Gao (2016). The optimal coordination, power flow
information on demand and supply gaps, UC and ED to minimize cost
between these MGs and distribution network is performed by DMPC.
function of MG are solved in optimal operation layer. Later, frequency
While EMPC is employed as LCs to minimize operating cost of whole
stability is achieved in frequency control layer. Although this gives
distribution system. Forecasting is used to manage uncertainties in
decoupled scheme for EMS and frequency control which increases DOF,
generation and demand. Though design is a bit complex, its benefits on
it has a complex structure. Similarly, in Berkel, Görges, and Liu (2013),
multi-MG system are substantial. MAS based controller implemented
a two level EMS using MPC is given. Lower layer works with a short
using modified diffusion communication scheme (called nudge me-
time scale (in secs) implements load frequency control (LFC), and upper
chanism) and optimal control principle is given in
one having large time interval (minutes) realizes ED and UC. The
Azevedo, Cintuglu, Ma, and Mohammed (2017). The two-level struc-
hierarchical MPC problem is formulated using a MILP, given in (15).
ture with lower one meant for realizing an adaptive droop strategy to
Advantage of this controller is that it includes charging/discharging
regulate frequency and upper one is for implementing penalty-based
along with mobility of EVs using an optimization scheme.
objective function subject to the generation limits for optimal MG
management. The proposed strategy is compared with the consensus Nu 1 NG 1 NEV
based algorithms and centralized control schemes. Simulation results min [JG, i (k ) + JSA, i (k )] + JT (k ) + JEV , i (k )
suggest it to be faster and utilize sparse information sharing. k=0 i=1 i=1 (15)
A cascade distributed secondary level control structure for power
management of dc MGs is proposed in Vu et al. (2017). This type of A hierarchical control system for the interfacing converters between
architecture avoids the mutual impacts of parallel controllers on each ac and dc buses of a hybrid MG is proposed in Lu et al. (2014). Here, the
other, which usually had degraded transient behavior. The proposed primary level implements decentralized PR based current control, PI
scheme has inner power control loop, while outer voltage loop. The based voltage and droop control for power sharing. Voltage deviation

137
S. Sen, V. Kumar Annual Reviews in Control 45 (2018) 118–151

arising from droop control is eliminated at secondary level by PI con-


troller. During grid-tied mode, power exchange between MG and ex-
ternal dc grids is controlled by tertiary level PI control. Design sim-
plicity makes it easy to use but, applicability is restricted to interface
hybrid MGs. In Che and Shahidehpour (2014), standard three-layered
architecture for economic and resilient operation of dc MGs is given.
Primary layer mimics the droop behaviour of ac MGs for power sharing
of among DERs using V-I droop, which is the main advantage of this
scheme. Secondary layer eliminates the voltage deviations by gen-
erating compensating voltage signal without modifying DER dispatch.
And, tertiary layer manages grid interactions and optimal dispatch of
DERs. A two-level master/slave structure for EMS is presented in Cai
and Mitra (2014). The upper one consists of MAS based power balance
controller, meant for maintaining equal demand-supply of P-Q. It uses a
fast three-sweep algorithm for real time operation. This layer also has
an ED, which provides control signals for slave controllers based on
comparison of incremental generation cost with neighboring DERs
generation agents. Here, the Lagrange function for ED is used as given
by (16), helps to include many objectives at a time, achieving optimi-
zation of whole MG. The lower layer follows the set points provided by
master layer.
n n n
L= Fi + PD Pi + i (P min + si2 Pi )
i Fig. 12. Hierarchical outage management scheme.
i=1 i=1 i=1
n

that power losses are minimized and enhanced voltage profile is ob-
max
+ i (Pi ti2 Pi )
i=1 (16) tained. Objective function optimization problem is solved through a
A three-level control of an energy hub having multi-source multi- self-adaptive GA and nonlinear programming method combined. Re-
product (MSMP) viz. natural gas, thermal and electrical energy is pro- ference signals are sent to LC of individual DERs to achieve optimal
posed in Xu, Jia, Wang, Yu, and Chiang (2015). The hierarchical EMS operation and reduced cost. Results on IEEE 33 bus system shows its
has a control layer for each purpose: supervision, optimizing and ex- better performance when compared to traditional dispatch, however,
ecution. The optimizing layer is further sub-divided into slow, medium here complexity is increased.
and fast for multi time-scale characteristic of the energy hub to ac- A two-stage hierarchical scheme for outage management to have an
commodate the dynamics of thermal, gas and electricity. Though dif- enhanced resiliency for a multi-MGs system is given in
ferent MSs and loads were controlled, it involves several implementa- Farzin, Firuzabad, and Moeini-Aghtaie (2016). Here, first stage consists
tion difficulties in terms of variable generation capacities, rating and of a MPC strategy for MG scheduling during any utility outage (see
dynamic characteristics. Fig. 12). At second stage, unused resilient resources of each MGs are
A control system with optimization of dc MGs operation by SoC coordinated by DSO for facilitating power transfer to unserved loads.
balancing of BESSs and enhancing overall efficiency is discussed in Optimization problem is formulated on MILP, integrating tie-line cost
Meng, Dragicevic, Vasquez, Guerrero, and Sanseverino (2015). In its for load lost risk. This EMS design serves the need of increased relia-
lower layer a primary droop controller is used, while at intermediate bility of MGs hence, has scope for realizing the smart grid features. The
level, secondary voltage restoration using PI control is implemented. At same authors extended this work further to propose a stochastic EMS to
tertiary level, objective function is formulated incorporating the feeder accommodate the unscheduled islanding events in Farzin, Firuzabad,
power loss and SoC deviations and solved by genetic algorithm (GA) at and Moeini-Aghtaie (2017). It gives an optimal scheduling of different
MGCC to have optimal load sharing. Virtual resistance of the DERs are MSs under prevailing uncertainties. Reduced risk of load loss and
considered as actual decision making variables, giving the benefit of simplified resynchronization with the main grid are some of its im-
improved power quality. In Sanseverino et al. (2015), the tertiary mediate benefits. Similarly, two level power management and co-
control layer is further divided into two sub-levels: EMS and tertiary ordinated operation of grid-tied dc MGs is proposed in Yang et al.
optimal power flow regulation. The EMS is designed using a linear cost (2016). The primary layer used SoC based droop control for the BESSs
minimization objective function subject to demand-supply balance interfacing dc-dc converters to achieve dynamic power sharing. Along
constraints, which is solved by using MILP. It provides inputs for with that, the V-P droop is implemented at PCC for bidirectional power
scheduling of DERs, which is passed on to lower sub-level optimizer by exchange. The secondary layer uses voltage based droop shifting to
EMS. Rolling horizontal method is used to solve optimal power flow restore bus voltage within rated value and SoC droop charging/dis-
problem, which gives an error free steady-state response. charging balancing to have optimal system operation. Slower dynamics
A three-layer MG control strategy is presented in Zhang, Liu, Zhao, and applicability to dc MGs only are its limitations. A bi-layer opera-
and Wang (2015). Here, local layer realizes a voltage regulator based tional optimization between DSO and MGs is proposed in
on an adaptive 3rd order sliding-mode direct voltage loop and virtual Bahramara, Moghaddam, and Haghifam (2016). Upper leader layer
negative impedance droop based power control. Middle cooperative maximizes DSO profit, while lower follower layer has to minimize
layer is meant for steady-state V-f control based upon PI method, also overall MG cost. The formulated bi-level nonlinear model is converted
accomplishing the phase synchronization to initiate grid connection. into single-level linear one using Karush–Kuhn–Tucker (KKT) criteria
Highest scheduling layer provides power exchanges. Although issues of and dual theory. This ascertains robust and stable characteristics of the
increased computational burden is present, the adaptive and fast controller by allowing easier computation of set points. A two level EMS
switching control action enables improved transient responses. It also for a cluster of MGs in distribution network is proposed in Lu et al.
helps to achieve seamless mode transitions. Lv, Ai, and Zhao (2015) 2017. The upper layer is meant for reducing the power losses, voltage
proposes a multi-purpose bi-level optimal coordination of grid-tied deviation and to minimize fluctuations in power sharing. While mini-
MGs. Upper level controller determines optimal dispatch of utility so mization of operating cost, emission in MGs are objectives of lower

138
Table 5
Summary of energy management system (EMS) techniques for microgrids.
Design technique Authors Control principle Contribution Comments (Benefits/limitations)
S. Sen, V. Kumar

Centralized control (Zaidi & Kupzog, 2008) Self-configurable control nodes First work on use of MGCC for load curtailment during Control nodes act as intelligent local controller, easy to realize.
peak hours.
(Tsikalakis & Hatziargyriou, 2008) Optimal control Economic optimization and market participation with Demand side bidding, minimization of operational cost and
upstream grid. maximizing profit.
(Ding et al., 2010) Optimal control An optimization and dispatch algorithm for MG. Environmental benefits, complex and slow process, sub-optimal
solution.
(Olivares et al., 2014) MILP and MPC An EMS that converts a nonlinear problem to simple Simplification is achieved without compromising much upon
linear and optimal control problem. accuracy.
(Ambia et al., 2014) PI controller Central supervisory Control (CSC) for power management Faster response, smoothens out power fluctuations.
of hybrid MG.
(Hajar et al., 2016) MPC Global management controller (GMC) for distribution Includes of forecasts information, but, overall design is complex.
system.
(Almada et al., 2016) Heuristic algorithm A heuristic master-slave architecture for EMS of ac MGs. Includes uncertainties of RESs, varying electricity tariffs along the
day.
Decentralized control (Wu et al., 2015) PI/PID control EMS for coordinated operation of RESs and BESSs using Has advantages of bus-signaling and droop control methods, but,
smooth switching droop control (SSDC). fails in economic optimization.
(Karavas et al., 2015) Fuzzy cognitive maps (FCM) and A multi-agent based EMS for a MG with various types of Comparatively economical, overall optimization is achieved.
MAS RESs.
(Li et al., 2016) MAS Decoupled EMS, separate electrical distribution and Easy to operate but, lacks coordination to facilitate market
communication. participation.
(Xia et al., 2016) PI control Unified multi-purpose controller for dc MG under both Provides stable design and immune to communication failures,
operating modes. time-delays.
(Sakurama & Miura, 2017) Decentral optimization Algorithm A demand response management algorithm using smart Optimization is done using Lagrangian algorithm, overall design
meters communication network topology. is highly complex.
(Nutkani et al., 2017) Cost-prioritized Optimization Comprehensive ED for islanded MGs with minimum Facilitates step-load changes, easy to realize, possible to set

139
online reserve. online reserve.
(Karimi et al., 2017) Modified P-f droop and PI control Load sharing and power management scheme for hybrid Uses frequency level triggering hence, reduced control effort,
PV/battery MGs. more accuracy.
(Baghaee et al., 2017) Sliding mode control (SMC) A multi-functional robust strategy for MG with nonlinear/ Independent of MG topology and load characteristics, faster
unbalanced load. response.
Distributed control (Costabeber et al., 2011) Optimal control Controller for PnP of RESs and reduction in power Uses PLC for communication, not Considered to be cost effective.
distribution loss.
(Nutkani et al., 2012) Droop control A power management between interlinked MGs by an Minimize power loss and simultaneously supports reactive power.
optimal number of interlinking inverters.
(Xu et al., 2013) MPC An optimal EMS achieving minimum excess/insufficient Guaranteed closed-loop stability, uses a one-step delay as in
energy in a MG. actual scenario.
(Shafiee et al., 2014) Consensus and adaptive control Dynamic averaging based power flow control among Extended feature of optimal storage, but, doesn't facilitate grid
cluster of dc MGs. interaction.
(Hossain et al., 2014; Hossain et al., 2016) H2 and LMI Controller to minimize negative interactions among Optimal response is obtained, complex implementation.
multiple MGs, and keeping optimal power flow.
(Li et al., 2015) PI control Optimization of interconnected-MGs to minimize the Controller design is simple, forecasting improves practical MG
operating cost. scenario.
(Wang et al., 2015; Wang et al., 2016) Sub-gradient algorithm A control of active power sharing between MG and utility Equal incremental rate attains minimum generation cost, faster
grid. control action.
(Zaery et al., 2015) Cooperative control Controller based on instantaneous generation cost to Produces of more power by cheaper MSs, prior knowledge of MSs
minimize the power generation cost in dc MGs. number is not required.
(Zhaoa & Dorfler, 2015) PI control A multi-objective controller for dc MGs to achieve droop, Mitigates voltage drift, optimal power injections and ED.
load sharing.
(Liu et al., 2015) MAS Load shedding of MGs by distributed cost optimization Optimal load shedding, considers overall and marginal costs,
(DCO) technique. facilitates PnP.
(Hamidia et al., 2015) Cooperative control Multi-purpose controller: V-f control, P-Q sharing, grid Working as EMS is not explored beyond just reconnection and
synchronization. synchronization.
(Stadler et al., 2016) MPC Control strategy for coordinated operation of DERs. Provides day-ahead load serving and peak saving, better ancillary
services.
(continued on next page)
Annual Reviews in Control 45 (2018) 118–151
Table 5 (continued)

Design technique Authors Control principle Contribution Comments (Benefits/limitations)

(Zaery et al., 2016) Consensus control Economic dispatch problem (EDP) and voltage regulation Ensures a lower total generation cost (TGC) and enhanced
S. Sen, V. Kumar

for a dc MG. stability, PnP.


(Xiao et al., 2016) Modified droop control Economic operation of a power park having several MGs Reliable, more economical and has faster response than
using frequency/incremental cost (f/IC). conventional droop.
(Mojica-Nava et al., 2016) Replicator dynamics controller A novel EMS based upon dynamic population game Considers feeders power losses, but, implementation difficulties
theory and heuristic search algorithm. in real scenario.
Distributed control (Li et al., 2016) MAS and consensus control A strategy that combines small time-scale AGC and large More reliable as (n-1) rule of used, applicable to both modes.
time-scale ED.
(Chen et al., 2016) Consensus control Secondary controller for optimal power sharing and Highly efficient analytical solutions on power flow, restricted
transmission loss reduction for dc MG. application to dc MGs only.
(Xu et al., 2016) MPC and receding-horizon ED problem of multi-MGs with PV-storage combination. Time-sharing price is used to fix time, rate of power supply to
optimization meet instantaneous load demand.
(Dorfler et al., 2016) Averaging PI control A distributed architecture for ED at tertiary level and More DOF for EMS and discrepancies are well compensated.
droop control.
(Kou et al., 2016) MPC New EMS for multi-MGs based upon distributed MPC Forecasting is used to manage uncertainties, however, it has a
(DMPC) and economic MPC (EMPC). complex design structure.
(Azevedo et al., 2017) MAS and optimal control Optimized MG operation by adaptive droop with penalty Faster response and utilizes sparse information sharing.
generation limits
(Vu et al., 2017) PI controller A cascade distributed secondary level control structure for Improved stability, better transient dynamics.
power management of dc MGs.
(Bersani et al., 2017) Robust and cooperative Control The problem of power exchange among IMGs is tackled. Needs partially nested information, controls maximum power
deviation.
(Li et al., 2018) Consensus averaging algorithm and Control and management of islanded MGs using dynamic During iterations the system balance (obtained by primary) is
MAS weights. never broken.
(Lin et al., 2018) Incremental cost droop and A controller for hybrid MGs to attain global system Relative loading index (RLI) to extract any hidden loading
canonical based control economic operation. conditions.the day.

140
Hierarchical control (Gil & Lopes, 2007) PI control Central autonomous management controller (CAMC) at Simple to realize, but, dynamic response is sluggish.
DSO level.
(Ding et al., 2013) MAS A three layered architecture for energy coordination of Usage of intelligent physical agents improves the speed of
isolated multi-MGs. operation.
(Nakazawa et al., 2013) Optimal control Operational control structure for MGs with frequency Increases DOF, RESs are considered as disturbances, complex
stability. design structure.
(Berkel et al., 2013) MPC and MILP A two-level EMS with lower level frequency control and Includes charging/discharging along with mobility of EVs for
upper level ED and unit commitment. optimization.
(Lu et al., 2014) PI and PR controllers Controller for interfacing converters between ac-dc buses Simple to design, only applicable to interfacing converter of
of hybrid MG. hybrid MG.
(Che & Shahidehpour, 2014) V-I droop Standard architecture for economic and resilient Eliminating voltage deviation without modifying DER.
operation of dc MGs.
(Cai & Mitra, 2014) MAS Two-level master/slave structure for EMS of islanded Achieves optimization of overall system, easy to realize.
MGs.
(Xu et al., 2015) Optimal control Control of energy hub with multi-source multi-product Implementation difficulties for variable generation capacities,
(MSMP). rating, dynamics.
(Meng et al., 2015) PI control tuned with genetic A hierarchical control system with optimization of dc MGs Obtain optimal load sharing, improved power quality.
algorithm (GA) operation by SoC balancing of BESSs.
(Sanseverino et al., 2015) MILP and rolling horizontal method Tertiary control layer is into a two sub-levels: EMS and Error free steady-state response, attains better optimization.
tertiary optimal power flow regulation.
(Zhang et al., 2015) PI and Sliding mode control (SMC) Third order control strategy for voltage regulation and Improved transient responses, seamless mode transitions, but,
droop based power control. increased computational burden.
(Lv et al., 2015) Self-adaptive GA and nonlinear A multi-purpose bi-level optimal coordinated operation of Better performance than traditional ED, however, complexity is
programming method. grid-tied MG and upstream network. also higher.
(Farzin et al., 2016;Farzin, Firuzabad, & MPC and MILP A two-stage scheme for outage management to have an Facilitates power transfer to unserved loads, smart grid features
Moeini-Aghtaie, 2017) enhanced resiliency for multi-MGs system. can be added.
(Yang et al., 2016) PI controller Power management and coordinated operation of grid- Droop shifting to restore bus voltage, slow dynamics, applicable
tied dc MGs. to dc MGs.
(continued on next page)
Annual Reviews in Control 45 (2018) 118–151
S. Sen, V. Kumar Annual Reviews in Control 45 (2018) 118–151

level. Based on game theory, interactions among the MGs and utility

Cost effective and synergic participation model, however more


network is model by an interactive game matrix (IGM) and optimized

Minimizes deviations between planned energy exchange and


by GA. Although complex to implement, it serves an important issue of
lowering the emissions.

actual one, but, it doesn't capture whole MG dynamics.


Ascertains robust and stable characteristics, simplified

Lowers emissions, however, complex implementation.


In Cominesi, Farina, Giulioni, Picasso, and Scattolini (2018, a two-
layer controller for EMS of multiple sources, storage and non-pre-
dictable loads is given. Higher level works as optimizer with slow
timescales (15 mins) and relies on simplified global MG model. It de-
cides nominal operating condition of MG over a long time horizon. The
lower level functions at a sampling time of 1 minute using stochastic

Comments (Benefits/limitations)
MPC. It reduces deviations between planned energy exchange and ac-
tual, accounting fast-rate stochastic fluctuations. Issue in this control
scheme is in using a discretized simple model which doesn't capture the
dynamics of complete MG. A robust EMS considering static modelling
of system frequency is discussed in Rezaei, Ahmadi, Khazali, and

computation.
Guerrero (2018). It formulates an energy and reserve management

complex.
design on frequency dependent characteristics of DERs. Later, in-
formation gap decision theory (IGDT) is used to handle MG un-
certainties and whole architecture is converted into single level MILP

Two-layer controller for EMS of multiple sources, storage


model to solve over a 24 h scheduling horizon. Although this gives a

A robust EMS considering the static modelling of system


Bi-layer operational optimization between the DSO and
cost effective and synergic participation model for both functions and

A two level EMS for cluster of MGs in distribution


demand response provider, computational complexity is increased en-
ormously. Summary of EMS of MGs is presented in Table 5.

4.5. Miscellaneous function

4.5.1. Centralized control


4.5.1.1. Battery management system (BMS). After discussing various

and non-predictable loads.


important control functionalities of MGs, still there are several other
substantial areas where the aforementioned control techniques are
applied. In this section, these objectives are presented as the
miscellaneous functions within a MG, which categorized into same
Contribution

frequency.
four control schemes as done earlier. In Qian, Zhang, Lai, and Yu
network.

(2011), a battery management system (BMS) controlling circuitry for


MGs.

charge-equalization of each cell by using the estimates of SoC is given.


For seamless power transfer between BESS and feeders, a dual-buck
half-bridge bidirectional ac-dc converter is used along with quasi-PR
Information gap decision theory
Dual theory and Karush-Kuhn-

scheme combined by an admittance compensator. It has features to


enable precision power flow and is successfully able to dictate the
working of BMS and interfacing inverter.
Game theory and GA

(IGDT) and MILP


Control principle

Stochastic MPC

4.5.1.2. Integrating electric vehicles (EVs). In Abdelaziz, Shaaban, Farag,


Tucker criteria

and El-Saadany (2014), a multi-stage centralized control strategy to


accommodate high penetration of Electric Vehicles (EVs) in an islanded
MG is proposed. The MGCC is designed with 3-stage droop based
optimal power flow schemes which fulfill several objectives. Firstly, it
minimizes the load shedding during insufficient generation, thereafter,
minimizing MGs’ operating cost and maximize EVs usage. Two
operating cases i.e. MG with and without adequate generation were
studied to validate effectiveness and robustness of proposed controller.

4.5.1.3. Coordination of RESs and BESSs. A centralized controller for


(Bahramara et al., 2016)

coordinated operation of various components viz. RESs, BESSs and


(Cominesi et al., 2018)

loads in is given in Diaz, Luna, Vasquez, and Guerrero (2017). Control


(Rezaei et al., 2018)

system is mainly having three supervisory tasks: active power


(Lu et al., 2017)

interruption to avoid batteries overcharging, load shedding to prevent


deep discharging, SOC equalization. It has the benefit of optimal
Authors

scheduled connection algorithm for loads, which reduces connection/


disconnection cycles, leading to increased life span of batteries.
Diaz et al. (2017) uses a novel MAS based on robot soccer
Table 5 (continued)

coordination strategy. It defines in prior, different tactics for DERs,


Design technique

BESS to operate in a specified collaborative manner viz. power balance


and power curtailment among DERs, SoC equalization, load shedding,
when operated in different roles: grid forming and grid following.
Advantage of proposed scheme is that, the used holistic system
algorithm which works on a knowledge-based task assignment,

141
S. Sen, V. Kumar Annual Reviews in Control 45 (2018) 118–151

enabling priority wise coordination. This method is limited by complex overloading and facilitates PnP capabilities.
implementation architecture.
4.5.2.2. Coordination of RESs and ESSs. The coordination of several
4.5.1.4. Power quality improvement. In Wang, Blaabjerg, Chen, and DERs under various topologies: ring, meshed and arbitrary complex is
Guerrero (2011), a selective harmonic compensator is proposed, discussed in Cucuzzella, Incremona, and Ferrara (2017). The MG model
which removes ‘whack-a-mole’ effect from voltage waveform using of is developed by a simple incidence matrix considering all node
low-BW channel as the harmonic signals are transmitted through it after connection parameters, unknown load dynamics, system nonlinearities
performing Park transformation. PI based regulators were used to and modeling uncertainties. Thereafter, two control strategies based
compensate the harmonics. Along with that, LCs for DERs were upon suboptimal second and third order SMCs are presented. Its
developed for effective nonlinear load sharing among themselves. benefits are that it works on the locally available voltage information
Although this method is quite effective for the transient disturbances, and results of stability analysis establish that the proposed control
it doesn't give a smooth steady-state response. In Tan, So, Chu, and scheme has asymptotic behaviour.
Chen (2013), MPC based power quality improvement strategy is given.
It integrates an extended Kalman filter (EKF) to estimate grid frequency 4.5.2.3. Parallel operation of VSCs. A control strategy for operation of
and extract harmonic spectra from V-I waveforms. Different test MG with high penetration of VSCs is proposed in Ovalle, Ramos, Bacha,
conditions viz. load sharing, power factor correction, grid voltage sag Hably, and Rumeau (2015). It considers VSC and LC filter combination
and swells were studied and results suggests it is capable to handle wide as a current source. This controller has only the knowledge of
varieties of power quality issues of MGs. inductance and capacitance of filter, while all other line parameters
A master/slave principle based control scheme to enhance the are unknown. Also, it applies the instantaneous power theory by
power flow among the three phases of a MG is presented in measuring/calculating V-I using theoretical methods, making
Brandao et al. (2017). Central master controller located at PCC pro- computations easy. In Tavakoli, Negnevitsky, and Muttaqi (2017), a
cesses unbalanced P-Q power terms and directs local slave controllers at MPC based voltage control and discrete SMC based current control of
DERSs to proportionally share them. This method is also known as inverters interfaced MSs is proposed. A test system is built using three
conservative power theory (CPT). Simulation was performed on real MSs with variable RLC loads. Thereafter, a performance comparison of
urban weak LV distribution network having asymmetric and non-si- proposed controller with a benchmark robust servomechanism (RSM)
nusoidal phase voltages, results suggesting efficacy for smoothening controller is performed. Simulation results suggest that control scheme
persistent fluctuations. is able to perform at par with RSM and has fast transient response and
tracking, robustness against unbalanced load transients and short
4.5.1.5. Polygeneration. The microgrids designed for remote location circuit conditions.
having a cold weather and at coastal area or islands with less drinking
water facilities needs additional features viz. room heating, potable 4.5.2.4. Polygeneration. In Kyriakarakos, Piromalis, Dounis, Arvanitis,
water by desalination, natural gas and hydrogen fuels etc. These are and Papadakis (2013) a MAS intelligent demand side management for
called the cogeneration or polygeneration microgrids an autonomous type polygeneration MG topology is proposed. Each of
Jana, Ray, Majoumerd, Assadi, and De (2017). As it involves several the agents have a well described Percepts, Actions, Goals and
type of sources, therefore its controllers are can be suitably realized Environment (PAGE). Here, the environment is taken to be non-
using one of the four control approaches. In this regard (Bracco et al., deterministic and inaccessible i.e. using grey prediction algorithms
2015) gives a centralized EMS consisting of a dynamic optimization only complete structure is implemented. Under overloading, it is
model formalized to the University of Genova smart polygeneration successfully able to turn off loads progressively without collapsing,
microgrid (SPM) along with sustainable energy building (SEB) testbed. hence, operations near the rated limits is possible. Similarly, in
It uses forecast data from the SCADA system and later, real time Karavas et al. (2015) also, a MAS and fuzzy cognitive mapping based
scheduling of different sources is done using a MPC based algorithm. It EMS of polygeneration MG is given. Three different of commodities i.e.
involves detailed electrical grid and power losses, EVs carbon footprint hydrogen, water, electricity were produced and stored as per the
and economic assessment. Basic advantage of this method is the decentralized decision making principle. Later, its results are
inclusion of on/off status of the microsources which allows to compared with centralized polygeneration control structure which is
customize various optimization scenario by receding horizon technique. observed to be at par with it.

4.5.2. Decentralized control 4.5.3. Distributed control


4.5.2.1. Battery management system. In Xu, Xiao, Hu, Wang, and Lee 4.5.3.1. Battery management system. A two level control strategy for
(2017), control for hybrid energy storage system (HESS) i.e. battery and BESS is proposed in Zhang, Cai, He, and Liu (2014). LCs consists of
supercapacitor is proposed. The battery converter is regulated by a micro-bank modules (MBMs) which includes a cell bank, micro-BMS
high-pass filter droop (HPFD) while a virtual capacitor droop (VCD) is and a dc-dc converter while connected to central BMS by CAN bus. By
implemented for controlling the supercapacitor converter. The issues taking into account the effect of battery recovery because of its
like accurate dynamic power sharing, SoC balance violation, steady intermittent discharging, a self-reconfigurable controller is used to
state bus voltage restoration are handled successfully. Advantage of this provide an optimal scheduling by calculating SoCs of each battery in
method is that it not only ensures continuous operation of HESS but, real-time. Some of its benefits are: reduced volume of BMS, no
acts as an effective power buffer as the included supercapacitor has overcharge issue, improved discharging efficiency and lower idling
faster dynamics. Multi-agent scheme based nonlinear SMC for SoC time. In Li et al. (2014), SoC equalization of dc MG based on a MAS
balancing is given in Morstyn, Savkin, Hredzak, and Agelidis (2017). algorithm is presented. Each battery unit is regarded as an agent that
Depending on the net load, all batteries are either in charging or sets its own reference voltage based on an average consensus algorithm
discharging mode so, prevent circulating current by avoiding so as to equalize SoC. By doing so, system stability is better maintained
unnecessary charging/discharging cycles. Whenever the SoC of any as compared to controlling the droop gain. Same authors used this
battery reaches boundary layer around average of its neighbours’ SoC strategy in Li et al. (2017), on ac MGs where the frequency set point is
level, its nonlinear SMC switches to a linear strategy therefore, prevents scheduled instead of droop gain. Both these methods have faster
chattering. The SoC controller is later integrated with existing MAS transient response as compared to conventional droop scheme.
based secondary voltage regulation and current sharing control Cooperative control of BESS using a two subsystems scheme: one for
strategies. This achieves faster SoC balancing, avoids battery charging, another during discharging modes of battery units is given in

142
S. Sen, V. Kumar Annual Reviews in Control 45 (2018) 118–151

Hosseinimehr, Ghosh, and Shahnia (2016). It also accomplishes the active synchronization of MG with utility grid is presented in
tasks of P-Q sharing among BESS units. A SoC based droop scheme is Sun et al. (2016). The synchronization correction signals are
developed to avoid fast depletion of units during discharging operation. generated from small-signal stability analysis and later used by
During charging, a PI controller is used to limit the units from voltage controllers to be transmitted to leader DERs only via sparse
overcharging surplus power from RESs. The Q assistance is also communication links. Each DER unit shares data with its neighbours,
achieved by an adaptive virtual impedance (AVI) control. While this voltage at PCC is synchronized by consensus. Low cost, simple approach
is highly effective but, the control architecture is too complex. are its main advantages which are aided by stable operation.
A secondary level BMS for a dc MG based upon dc-bus signaling
(DBS) and compensating virtual droop impedance of BESSs is proposed 4.5.3.4. Prevention of cyber-attack on communication links. In (Lu et al.
in Oliveira, Silva, and Donoso-Garcia (2017). Depending upon magni- (2016), secondary level control strategy that mitigates effects of
tude of SoC disparity (between each storage unit and average SoC), malicious communication link attacks is discussed. The primary level
output current is generated to force charge equalization during both is of P-f droop and secondary level has consensus optimization
charging/discharging modes. Here, use of convergence factor is done algorithm which is updated by dual-ascent method. It also achieves
for speeding up the SoC balancing, and its tradeoffs with voltage de- nominal frequency and proper power sharing. Two attack models or
viation and interference are also discussed. A distributed control adversary inputs i.e. link- and node-based were tested upon MG
scheme for several PVs and BESS hybrid units is proposed in communication channel. Results show it is quite efficient to nullify
Golsorkhi et al. (2017). It is divided into two levels: primary level with link failures.
decentralized V-I droop and secondary level with coordinated leader-
follower method. The average MG voltage is restored by the leader, 4.5.3.5. Power quality improvement. A distributed consensus algorithm
while the follower balances SoC by managing power sharing among (DCA) for sharing of the negative sequence current is given in
BESS. The proposed scheme is simple and easy to implement on large Meng et al. (2016). The MSs are controlled to compensate for voltage
scale MGs but, it has slower dynamics. Voltage control of MGs using unbalances by performing a detailed unbalance analysis, which also
coordination of multiple electric springs (ESs) to mitigate uncertainty in improves overall power quality of the MG. The experimentation results
RESs is proposed in Chen, Hou, and Hui (2017). At first, a controller for suggest it has an enhanced system flexibility to execute the PnP feature
single ES unit is designed by PR principle. Thereafter, using information of incoming/outgoing DERs.
sharing among neighboring units, a consensus based algorithm is de-
veloped to assign reference voltages for multiple ESs. This method has 4.5.3.6. Stability enhancement. In Wu and Shen (2017), an optimal
advantages of achieving synchronization of new ESs unit because it uses controller at secondary level is proposed for stable and robust operation
of local variables only. Comparison conducted with conventional droop under various MG operating conditions. It helps in attaining desired
strategy shows its improved capabilities in maintaining voltage level of dynamic performances i.e. fast response, proper damping and stability
critical loads. On a similar note in Nasirian, Yadav, Lewis, and Davoudi margin. A complete stability analysis on the developed unified small-
(2017), a collective and fully distributive control for operation of power signal based dynamic model of a resistive MG is done to establish a
buffers using sparse communication for a dc MG is given. Each of the mechanism to mitigate the undesirable dynamics and oscillations at
critical load is interfaced via power buffer, which uses local measure- secondary layer. This helps to achieve increased stability margins under
ments to set a desired energy profile during transients. It also manages different scenarios.
energy reservoirs to supply high quality power to the grid.
4.5.3.7. Polygeneration. A dynamic market mechanism (DMM) for a
4.5.3.2. Coordination of RESs and BESSs. Seamless operation of DERs CHP microgrid based on distributed scheme is proposed in Nudell et al.
and participation in frequency response, power sharing in both grid- (2017). It uses the latest up-to-date estimates of thermal and electric
tied and islanded modes is presented in Karami, Seifi, and loads and the renewables hence, allows the PnP features of the BESSs,
Mohammadian (2016). It uses a droop scheme based on power angle RESs or a new smart building without reformulating the optimization
for power sharing among DERs during islanded operation and problem. The testbed at University of Genova is considered to obtain
supplying constant power under grid-tied, without the requirement of the market clearing price and set points for thermal and electric
any switching over from one mode to another. PnP capability and components along with non-binding estimates of these variables. The
stability analysis of the proposed controller is discussed using a computational complexity of the proposed method is found to be
developed state-space model. Auto mode adapting of the controller is reduced due to the usage convex optimization technique.
definitely an additional feature. Coordination of multiple PMSG based
wind turbines in a dc MG to maintain demand-supply balance and 4.5.3.8. Market participation. In Iggland, Wiget, Chatzivasileiadis, and
achieving optimal load sharing is given in Kou, Liang, and Gao (2017). Anderson (2015), an optimal power flow (OPF) for economically
Grid side converter (GSC) is controlled by two levels: upper one consists efficient generation dispatch in multi-area hybrid MGs i.e. HVAC and
of MPC approach to optimize parallel connected interfacing buck HVDC is proposed. This method gives a distributed optimization
converters in a cooperative way. It generates the reference power set algorithm using the linearized OPF formulation. It considers three
points, which are input for lower layer PMSG pitch angle controllers. operating scenario for the multi-area HVDC grid. Advantage of the
Advantage of this scheme is that along with controlling an ac proposed method lies in the fact that it uses less number of variables for
microsource in a dc MG, it achieves system-wide objectives. data exchange, and hence, reduces the amount of computations per
Distributed decision-making units that depend upon the knowledge iteration. Also, it removes the need for a system-wide reference bus,
based fuzzy inference systems for coordination of RESs and BESSs for is instead considers local reference bus for each area. It can support the
given in Diaz, Vasquez, and Guerrero (2018). The data sharing is determination of security constrained based OPF. Recently, in Li et al.
achieved via voltage-frequency bus-signaling scheme with neighboring (2018), a coordinated transmission and distribution ac optimal power
units to have a global level operational perspective. It considers various flow (TDOPF) is proposed. First, mathematical model of TDOPF is
stages, constraints to attain SoC balancing of batteries, achieves developed and later it is solved by a master-slave distributed structure
appropriate charging profiles and power curtailment for noncritical using heterogeneous decomposition algorithm. Later, this method is
loads. Even with these benefits, this design is highly complex and has compared to multi-area OPF and improvements are reported. It helps to
slow transient features. mitigate the voltage rises due to load curtailment, decreases mismatch
in boundary bus, improves the economic operation allows to
4.5.3.3. MG synchronization. A cooperative scheme for seamlessly accommodate more DERs by the grid.

143
S. Sen, V. Kumar Annual Reviews in Control 45 (2018) 118–151

4.5.4. Hierarchical control small fluctuations were not eliminated at steady-state.


4.5.4.1. Battery management system. A hierarchical controller for power A strategy to improve reliability during utility grid faults, which
sharing among various BESSs is proposed in Wang, Xiao, Setyawan, Jin, mitigates voltage sags and maintains power quality is proposed in
and Hoong (2014). During the normal working of communication links, Zhao et al. (2015). Its two layers are: lower one consists of a primary
centralized method of control is used which divides net power and droop and a virtual impedance loop. The upper layer consists of two
assigns proper power set points among BESSs depending upon SoCs. If controllers i.e. positive and negative sequence droop controllers to in-
in case a link fails, distributed control technique gets activated, which ject both of them, to achieve proper load sharing accuracy among DERs
use dc-bus signaling (DBS) to set the power scheduling in autonomous and mitigate voltage sags by providing ride through capability, elim-
way. Later, this work is further extended in Xiao, Wang, and Setyawan inating negative sequence circulating current among DERs respectively.
(2015), where in centralized layer, secondary voltage regulator is also Though voltage waveforms are free of harmonics, injecting sequence
utilized to minimize its deviation. Hardware verification is performed components reduces stability of an inverters. On a similar note, in
on a laboratory scale MG, where the effects of ramp rate of converters Lorzadeh, Savaghebi, Abyaneh, and Guerrero (2015), two level control
and the time delays in data acquisition were studied. Results show it is strategy for accurate Q sharing and harmonic current among VSIs is
able to successfully take care of these contingencies. given. Primary level extracts different harmonic frequencies and
A three layer EMS for a MG is proposed in Cominesi, Bella, Farina, transmits these to secondary level. Which later, computes the respective
and Scattolini (2016). Devices with similar dynamics viz. the dis- instantaneous circulating currents for generating control signals for
patchable and non-dispatchable sources, batteries, loads are clustered VSIs as voltage set points. Here, improved power quality is achieved
together so that the computational burden of highest layer that defines along with Q-V droop control simultaneously.
amount of energy exchange, is reduced. The middle supervisory layer A control scheme for suppressing unbalanced voltage and harmo-
assigns the energy demands for units in each group. The lower layer nics in grid-connected MGs is proposed in Feng, Sun, Guan, Guerrero,
with built-in local controllers extracts required energy in proportion to and Xiao (2017). A multiple PR based voltage regulating compensator is
units’ rating. It simultaneously manages power sharing with utility, integrated to secondary layer to obtain references for negative funda-
BMS with less burden on controller and other resources. A stochastic mental and unbalanced harmonic voltage. The primary layer con-
multi-objective MILP based optimal scheduling of BESS is discussed in stituted of PI and multiple PR controllers for tracking of VSI voltage.
Farzin et al. (2017). It considers normal operating cost of MG along Thereafter, stability analysis is performed and the magnitude of re-
with load curtailment index under unscheduled islanding events, while duced margins is found so that parameter set point values can be ob-
only the probabilistic distribution of its duration is required to be tained by using the developed system model. Simple to implement,
computed. These features helps to achieve a trade-off between the se- good response for both transient and steady-state conditions are some
curity and economic criteria for BESSs scheduling. Use of fuzzy decision of its benefits.
making approach at operator level also gives optimal performance With all these miscellaneous functions, main focus on the control
under grid operation. aspects of MG is completed. However, as the increasing demand for
electric transportation system, the microgrids can serve an integrated
4.5.4.2. Coordination of RES and BESSs. A unified cooperative PI role for aggregating the electric vehicles (EVs) into the LV distributed
control based strategy is presented in Moradi, Eskandari, and system. The MGs can mostly be used to realize the charging stations
Hosseinian (2015) for coordinated operation of RESs and BESSs in facility for the plug-in EVs. This would not only enhance the energy and
islanded mode of operation. Battery units are used to handle slow storage capacity of the MG but also, supports the balancing of excess or
response of MSs. Droop scheme at primary level for controlling the PE less power generated power by the RESs. As this work concentrates on
converters to improve voltage regulation and load sharing. The network MG control, the topic of EVs integration is not covered here. But, in-
is considered to be resistive and a master converter is designed as VSI, terested readers can go through the article discussed in the following. In
while rest of the RESs and battery converters are working as CSI type to Green, Wang, and Alam (2011), Morais, Sousa, Vale, and Faria (2014),
achieve V-f tracking and fast dynamic response for supplying the various impact of the EVs on the distribution network viz. vehicle pe-
required power into the MG network. netration, driving pattern, charging timing and characteristics, power
demand curve in smart grid environment etc. are discussed. This fea-
4.5.4.3. Market participation. A hierarchical control based demand tures give rise to concepts of vehicle-to-grid (V2G) and grid-to-vehicle
response technique is proposed in Bhattarai, Bak-Jensen, Mahat, (G2V). In García-Villalobos, Zamora, Martín, Asensio, and Aperribay
Pillai, and Maier (2013). First a brief discussion on classification of (2014), different smart charging control approaches are given. Model-
various demand response methods along with coordinative goals of ling, integration, control, optimization and fleet management of these
other actors viz. aggregator, the DSO are specified, thereafter a three EVs within the visions of smart grid opens up new areas for research
layered lower to higher progressive delays based demand response (Hu et al., 2016). Also, their economic and technical viability are need
activation time is given. This allows the control action to be liberal and to be studied. In this regard, Shafie-khah, Neyestani, Damavandi, Gil,
not precise as in the case of centralized demand response. However, this and Catalão (2016) presents various demand response management
framework is found to be not as accurate as its centralized counterpart. techniques for market participation of MGs in presence of interacting
RESs and the integrated EVs. Similarly, in Alipour, Mohammadi-
4.5.4.4. Power quality improvement. A two-layer hierarchical control Ivatloo, Moradi-Dalvand, and Zare (2017), stochastic optimal sche-
strategy for compensating voltage harmonics of MGs by coordinating duling of aggregators of EVs for the energy and ancillary service market
the DERs and active filters is discussed in Savaghebi, Hashempour, and under the uncertainties of vehicle availability, market prices and status
Guerrero (2014). The primary layer consists of droop controller to share in reserve market are discussed.
fundamental power among DERs, a virtual impedance loop to share
required harmonic currents and PR based V-I controllers for generating 5. Research challenges of microgrids control
output voltage of PWM gate fired inverters. The central secondary
controller compensates voltage harmonics of sensitive load bus (SLB) by As discussed earlier, despite of having huge benefits, the real world
initiating proper control signals to DERs and making the active power implementation of MGs concept is not at par with its research devel-
filters in participating for harmonic compensation. Same authors extend opments over the years. Also, the existing ones are yet far from having
this work further to multi-MGs level in attributes of an ideal MG. There are various obstacles which need to be
Hashempour, Savaghebi, Vasquez, and Guerrero (2015). In both identified and resolved. Off course these can be technical and non-
designs simple PR were used giving fast response, however, persistent technical viz. regulatory and financial etc. As this work is meant for

144
S. Sen, V. Kumar Annual Reviews in Control 45 (2018) 118–151

MGs control, the non-technical barriers are not discussed here. The sequence are required to be studied. Proper control scheme for
interested readers can find more on this topic in oscillation free active power and ripple free voltage need to be
Soshinskaya, Graus, Guerrero, and Vasquez (2014). Here, it is im- developed.
portant to recognize that there are some existing microgrid testbeds • Plug-and-play (PnP) capabilities: connecting MSs and other per-
around the world as discussed in Hossain et al. (2014), Lidula and ipherals without disturbing the MGs’ operational stability is not
Rajapakse (2011) Ustun, Ozansoy, and Zayegh (2011). The CERTS an easy task to be realized, it's limited by reconfigure require-
testbed in Ohio, USA is the one of the most researched MGs around the ments for real time synchronization.
world. It uses a peer-to-peer i.e. the decentralized control strategy using • Load shedding algorithms: to provide reliable supply for critical
P-f and Q-V droop with resynchronization features. Another popular loads, managing demands during the peak hours, needs a con-
MG system is available at National Technical University of Athens solidated design and implementation of load shedding algorithm
(NTUA), Greece where MAS based control technique is implemented for interruptible and shiftable loads.
using four agents viz. production unit, consumption unit, power system
and MGCC. A demonstrative project at Aichi, Japan is maintained by Various research gaps and challenges for MG control are suggested
the New Energy and Industrial Technology Development Organization in the earlier discussed four control architectures.
(NEDO) is used to reduce demand-supply gap with consideration of
day-ahead planning with GA and a meta-heuristic search technique. 1. Centralized control:
From these discussion, it can be said that, real scale uses of MGs is still • Optimal demand response: design of MGCC for achieving proper
in experimental stage and several outstanding technical issues are demand side management (DSM) under various fluctuating
needed to be resolved (Hirscha, Parag and Guerrero, 2018). This section quantities/variables i.e. generation, grid price, status of MSs and
briefly discusses various research gaps, technical challenges existing in loads, LCs’ parameters etc.
present MG control literature, resolving which would lead to more real • Multiple interlinking converters: a master/slave centralized archi-
scale implementation. tecture for P-Q exchange and V-f regulation for power quality
The technical obstacles are identified under the perspectives of improvement along with fault current limiters in a multi-MGs
global and local/component level for the MGs. scenario is need to be developed.
• Market participation: reduced operational cost could be achieved
1. Global/whole MG level issues: by including market prices, supply and demand bids along with
• Islanding detection and initiation: is the most important feature of short term load-generation forecasting into the EMS algorithms
MGs, islanding detection within minimum time of a utility fault while dealing with the upstream DSO for providing ancillary
and initiating various adjustments at each component (RESs, services.
loads, PE converters) level, such that the reliability of MG system • Robust decision making: use of fuzzy logic, MPC and IMC instead of
is maintained. existing classical multiple control loops are open for research and
• Resynchronization with upstream network: after having the fault a performance assessment for identifying best suited for uncertain
clearance, exact moment of reclosing the switch/CB at PCC, calls MG systems.
for a substantial amount of control functions i.e. regulate V-f and • Protection functions of MGCC: can accommodate various adaptive
minimizing the load and generation power mismatch. relay functions, differential schemes, variability of DERs, power
• Maintaining power quality: due to use of fluctuating RESs the DSO rate limits, dual-mode operation etc. needs more exposition.
during grid-tied has to maintain nominal V-f and P-Q balance, 2. Decentralized control:
which is difficult because of coupling nature of P-Q due to high R/ • Detailed documentation of droop variants: a comparative study and
X ratio of LV network. performance assessment of various droop control types to find the
• Coordinated control of harmonics: extensive presence of PE con- best suitable for real scale application on the LV MG networks.
verters causes high risk of circulating current and local over- • Cost-prioritized nonlinear droop: to simultaneously attain optimal
voltage at DERs ends. Proper techniques are needed to be devel- and fast transient response at primary layer, a cost- based objec-
oped to dealt with these problems. tive function with nonlinear droop coefficient updation is re-
• Protection from internal faults: during external/main grid faults, quired to be developed.
MGs’ islanding is a solution. But, for internal faults, the inverter • Power line or dc-bus signaling: certain amount of global coordina-
based MSs doesn't generate the rated amount of fault currents tion can be achieved by DBS communication by inducing voltage
required for detection and tripping of the conventional over- level changes at control nodes by the LCs for each device. This
current relays. area needs a thorough research.
• Planning of new MGs: while building a MG at a remote location the • System inertia: as inverter based MGs lacks inertial traits of con-
planning plays an important role. Proximity of ac and dc MSs and ventional power system, each controller designed for PEI of DERs
loads should be carefully planned to have an optimal design with should generate certain magnitude of virtual inertia to keep nor-
least power losses. malcy of MG frequency.
2. Component level/distributed issues: 3. Distributed control:
• High maintenance components: fuel cell, lead-acid battery systems • Stability of overall system: though this enhances reliability and
etc. calls for regular maintenance. Some personnel have to be redundancy in comparison to centralized scheme, more under-
deployed for the purpose. standing on their effects on convergence and stability of the
• Minimizing generation fluctuations: weather dependence of RESs overall MG system is need to be studied.
always creates major problems in managing the demand-supply • Model-free based algorithms: control strategies which are in-
balance. Usage of storage devices with the adjustable maximum dependent of detailed MG model and robust to changes in system
power point tracking (MPPT) would play a supporting role. parameters are required to be developed to suit the inherently
• Communication between agents: multi-agent based control archi- uncertain MG operating scenario.
tecture requires regular contact and proper negotiation algorithms • Interaction of multi-agents in multi-MG system: networked MGs with
for a coordinated operation, are often difficult to be implement for interlinking agents/controllers are required to be analyzed for
prolific components of MGs. negotiating amount of power sharing between themselves, while
• Unbalanced grid conditions: working of DERs to maintain power maintaining proper P-Q level within.
quality during the unbalanced conditions with un-symmetric V-I • Cybersecurity considerations: developing robust control schemes to

145
S. Sen, V. Kumar Annual Reviews in Control 45 (2018) 118–151

prevent cyber-attacks on communication links requires more at- interference and high data transfer rate, optical fibers are con-
tention. sidered most reliable medium.
• Pinning control: with advantages of being computationally efficient 2. Measurement/Monitoring infrastructure:
and smaller number of controller/agents usage than conventional • Smart metering infrastructure (SMI): it facilitates two-way com-
MAS, more emphasis is need to be given on its application. munication, records electricity usage at hourly interval and sends
4. Hierarchical control: the data to the utility grid central management system for mon-
• Unified control scheme for mode switching: smooth mode transition itoring and billing. It also receives the information on unit elec-
i.e. islanding and resynchronization of MG needs assistance from a tricity price at each hour, so that, customer may adjust its con-
single circuit configuration with unified design to be able to in- sumption.
stantaneously shift its role to and fro the islanded and grid-con- • Supervisory control and data acquisition (SCADA) system: works on
nected operation. a wide area and supports the DSOs in decision making. Facilitates
• Closed loop designs: most hierarchical structure are open loop de- data acquisition from important buses, supervisory control and
sign: primary-secondary-tertiary, while for better optimization activates alarms during critical operating scenarios. Coordinates
and efficient operation, feedback signals from upper to lower between EMS and RTUs.
layer are required for set points selection. • Phasor measurement units (PMU): though costly, but most efficient
• Cooperative power quality control: the inclusion of EVs in MGs technology for the wide-area measurement system (WAMS),
systems the huge charging current leads to deteriorate power readily provides the voltage, current magnitudes and phase angles
quality, need for integrated control for improving voltage fluc- with precise time synchronization. Must be implemented for
tuation and harmonics compensation. multi-MGs optimization and control.
• Modern control techniques: classical PI control is the most exploited • Advanced forecasting techniques: load demand, generation and
one, while multi-layer optimal, MPC are not yet been explored. A electricity price forecasting plays a crucial role in MG concepts. It
detailed research on implementation of these on MGs hierarchical provides statistics based pseudo measurement, for deciding stra-
functions is open for research. tegies for next 24/48 hours. It would also accommodate various
• Outage management scheme: to maintain high reliability, especially RESs and load uncertainties.
for critical loads and coordination between rest of the loads, a 3. Control infrastructure:
proper OMS is an absolute requirement. • Remote terminal units (RTU): can act as local controllers (LCs) at
each component of a MG. These micro-controllers based special
6. Role of microgrids in realizing smart grid purpose computers receives set points/status as digital input and
generates digital control output.
This section presents the role of MGs for realizing smart grid tech- • Distributed control system (DCS): due to distributed nature of var-
nologies. It briefly discusses several basic integrated technologies and ious entities in a MG, a complete distributed control infrastructure
brings various future trends of MGs concepts to fulfill the requirements is inevitable. It achieves global optimization, reconfigure the LCs
of a smart grid. It can be defined as electric power system where all the and facilitates PnP capabilities.
functions viz. generation, delivery (transmission and distribution), • Energy management system (EMS): central unit to manage MG and
storage and end usage are all automated and cyber enabled utility grid interaction, maintain demand-supply balance and
(Annaswamy, 2013). The important concepts of system modeling and provide redundant backup for all the devices. It optimizes the MG
identification, state estimation, optimization, adaptive and feedback operation with ED and UC.
control etc. will be key enablers. This electricity infrastructure would • Vehicle to grid (V2G) technology: could facilitate spinning reserve,
have the following traits: load balancing, improve efficiency and reliability by charging/
discharging of EVs as per need.
• Integrates intermittent RESs to achieve decarbonization; 4. Protection infrastructure:
• Facilitate two-way exchange of power and information; • Protective (directional over current) relays: these smart, remotely
• Optimized resource/asset utilization; reconfigurable, minimum time relays are primary requirements
• Provides reliable power efficiently, with flexible demand manage- for fault isolation, equipment protection and preserving MGs’
ment and customer choices; stability. These should be able to handle the bidirectional fault
• Possess self-healing capabilities from any disturbance event in the current.
network; • Programmable logic controller (PLC): real time extended input/
• Be resilient to physical and cyber-attacks. output (I/O) functions of various relays and control architecture
asks for use of PLCs. It incorporates adaptive, intelligent and
The power system is always been considered to be “smart” at the backup features into MGs technologies.
transmission level. After the deregulated restructuring, distribution • Static transfer switch (STS): enhances reliability of critical loads by
network is presently evolving and with inception of MGs concepts it is allowing instantaneous transfer of power source, also provides
becoming more smart. Some of the devices for smart grid functions are ride-through capability during short duration power disturbances.
given in Yoldas, Onen, Muyeen, Vasilakos, and Alan (2017). Further • Islanding detection methods (IDM): essential requirement for
contributions towards it could be availed by integrating the following maintain reliability and stability of MG system during utility grid
into MG technologies: faults, by getting separated within time at PCC. Several methods
are still under research.
1. Communication infrastructure:
• Power line carrier communication (PLCC): uses existing distribution 7. Conclusions
lines, hence provides a low cost information exchange medium for
transmitting the modulated signals at radio frequency between MGs, a cluster of various electrical and electronic devices, are re-
10 kHz to 490 kHz. garded as the best possible way to realize the smart grid features. This
• Digital subscriber lines (DSL): another low cost option for high work compiles various control aspects of MGs viz. its functional re-
speed communication is by using existing, widespread telephone quirements, architectures, level structuring, classification. Due to its
line infrastructure. inherent SoSs perspective, the control structures are implemented
• Optical ground wire (OPGW): with low latency, immunity to under these four heads: centralized, decentralized, distributed and

146
S. Sen, V. Kumar Annual Reviews in Control 45 (2018) 118–151

hierarchical. A comprehensive survey of MGs’ droop control, V-f reg- Bidram, A., Davoudi, A., & Lewis, F. L. (2014b). A multi-objective distributed control
ulation, P-Q sharing, EMS are documented. Here, an extensive list of framework for islanded ac microgrids. IEEE Transactions on Industrial Informatics, 10
(3), 1785– 1798.
references was compiled. Several miscellaneous applications i.e. BMS, Bolognani, S., & Zampieri, S. (2011). Distributed control for optimal reactive power
power quality improvement, coordinated parallel operation of RESs and compensation in smart microgrids. Proceedings of the fiftieth IEEE conference on de-
BESSs etc. are also reported. Thereafter, several technical barriers/ob- cision and control and European control conference (pp. 6630–6635). .
Bolognani, S., & Zampieri, S. (2013). A distributed control strategy for reactive power
stacles which prevent real scale usage of MGs were identified and other compensation in smart microgrids. IEEE Transactions on Automatic Control, 58 (11),
research gaps on MG control functions are suggested for SG researchers. 2818– 2833.
Finally, the role of MG and other integrated technological improve- Bracco, S., Delfino, F., Pampararo, F., Robba, M., & Rossi, M. (2015). A dynamic
optimization-based architecture for polygeneration microgrids with tri-generation,
ments required for realizing future SG were briefly summarized. The renewables, storage systems and electrical vehicles. Energy Conversion and
budding researchers in SG field would acknowledge this paper for Management, 96, 511– 520.
bringing different control aspects on MGs under one cover. Brandao, D., Caldognetto, T., Marafao, F., Simoes, M., Pomilio, J., & Tenti, P. (2017).
Centralized control of distributed single-phase inverters arbitrarily connected to
three-phase four-wire microgrids. IEEE Transactions on Smart Grid, 8 (1), 437– 446.
References Cady, S. T., Domínguez-García, A. D., & Hadjicostis, C. N. (2015a). A distributed
generation control architecture for islanded ac microgrids. IEEE Transactions on
Abdelaziz, M., Shaaban, M., Farag, H., & El-Saadany, E. (2014). A multistage cen- Control Systems Technology, 23 (5), 1717– 1735.
tralized control scheme for islanded microgrids with PEVs. IEEE Transactions on Cai, H., & Hu, G. (2018). Distributed robust hierarchical power sharing control of grid-
Sustainable Energy, 5 (3), 927– 937. connected spatially concentrated ac microgrid. IEEE Transactions on Control Systems
Ahmadi, S., Nazarpour, D., Shafiee, Q., & Bevrani, H. (2016). A fuzzy inference model Technology In press.
for distributed secondary control of islanded microgrids. Proceedings of the twenty Cai, N., & Mitra, J. (2014). A multi-level control architecture for master-slave organized
fourth Iranian conference on electrical engineering (pp. 1233–1238). . microgrids with power electronic interfaces. Electric Power Systems Research, 109,
Ahn, C., & Peng, H. (2013). Decentralized voltage control to minimize distribution 8– 19.
power loss of microgrids. IEEE Transactions on Smart Grid, 4 (3), 1297– 1304. Cai, H., Hu, G., Lewis, F. L., & Davoudi, A. (2016). A distributed feedforward approach
Akbari, M., Golkar, M. A., & Tafreshi, S. M. M. (2011). Voltage control of a hybrid ac/dc to cooperative control of ac microgrids. IEEE Transactions on Power Systems, 31 (5),
microgrid in grid-connected operation mode. IEEE Proceedings of the PES innovative 4057– 4067.
smart grid technologies (pp. 358–362). . Caldognetto, T., Tenti, P., Mattavelli, P., Buso, S., & Brandao, D. I. (2015). Cooperative
Alipour, M., Mohammadi-Ivatloo, B., Moradi-Dalvand, M., & Zare, K. (2017). Stochastic compensation of unwanted current terms in low-voltage microgrids by distributed
scheduling of aggregators of plug-in electric vehicles for participation in energy and power-based control. Proceedings of the IEEE thirteenth Brazilian PE conference and first
ancillary service markets. Energy, 118, 1168– 1179. southern PE conference (pp. 1–7). .
Almada, J. B., Leao, R. P., Ampaio, R. F., & Barroso, G. C. (2016). A centralized and Canizares, C. A. (2014). Trends in Microgrid Control. IEEE Transactions on Smart Grid, 5
heuristic approach for energy management of an ac microgrid. Renewable and (4), 1905– 1919.
Sustainable Energy Reviews, 60, 1396– 1404. Cavraro, G., Caldognetto, T., Carli, R., & Tenti, P. (2016). A master/slave control of
Ambia, M., Al-Durra, A., Caruana, C., & Muyeen, S. (2014). Power management of distributed energy resources in low-voltage microgrids. Proceedings of the European
hybrid micro-grid system by a generic centralized supervisory control scheme. control conference (pp. 1507–1512). .
Sustainable Energy Technologies and Assessments, 8, 57– 65. Chamorro, H. R., & Díaz, N. L. (2013). Hierarchical power flow control in low voltage
Anand, S., Fernandes, B., & Guerrero, J. (2013). Distributed control to ensure pro- microgrids. Proceedings of the North American power symposium (pp. 1–5). .
portional load sharing and improve voltage regulation in low-voltage dc Microgrids. Chang, C., & Zhang, W. (2015). On distributed control of voltage source inverters in
IEEE Transactions on Power Electronics, 28 (4), 1900– 1913. island ac microgrids. Proceedings of the American control conference (pp. 2365–
Annaswamy, A. M (Ed.). (2013). IEEE vision for smart grid controls: 2030 and beyond IEEE 2370). .
Press Publications . Chang, C. Y., & Zhang, W. (2016). Distributed control of inverter-based lossy microgrids
Asad, R., & Kazemi, A. (2014). A novel distributed optimal power sharing method for for power sharing and frequency regulation under voltage constraints. Automatica,
radial dc microgrids with different distributed energy sources. Energy, 72, 291– 299. 66, 85– 95.
Avila, N. F., & Chu, C. (2016). Distributed pinning droop control in isolated ac micro- Che, L., & Shahidehpour, M. (2014). DC microgrids: Economic operation and en-
grids. Proceedings of the American control conference IEEE industry applications society hancement of resilience by hierarchical control. IEEE Transaction on Smart Grid, 5
annual meeting (pp. 1–8). . (5), 2517– 2526.
Azevedo, R., Cintuglu, M. H., Ma, T., & Mohammed, O. A. (2017). Multiagent-based Chen, G., & Feng, E. (2015). Distributed secondary control and optimal power sharing in
optimal microgrid control using fully distributed diffusion strategy. IEEE Transactions microgrids. IEEE/CAA J. of Automatica Sinica, 2 (3), 304– 312.
on Smart Grid, 8 (4), 1997– 2008. Chen, F., Burgos, R., Boroyevich, D., Rodriguez-Diaz, E., Meng, L., Vasquez, J. C., et al.
Baghaee, H. R., Mirsalim, M., & Gharehpetian, G. B. (2016). Power calculation using (2016). Analysis and distributed control of power flow in dc microgrids to improve
RBF neural networks to improve power sharing of hierarchical control scheme in system efficiency. Proceedings of the fourth international symposium on environmental
multi-der microgrids. IEEE Journal of Emerging and Selected Topics in PE, 4 (4), 1217– friendly energies and applications (pp. 1–6). .
1225. Chen, X., Hou, Y., & Hui, S. Y. (2017). Distributed control of multiple electric springs for
Baghaee, H. R., Mirsalim, M., Gharehpetian, G. B., & Talebi, H. A. (2017). A decen- voltage control in microgrid. IEEE Transactions on Smart Grid, 8 (3), 1350– 1359.
tralized power management and sliding mode control strategy for hybrid ac/dc mi- Cominesi, S. R., Bella, A., Farina, M., & Scattolini, R. (2016). A multi-layer control
crogrids including renewable energy resources. IEEE Transactions on Industrial scheme for microgrid energy management. IFAC Conference, 49 (27), 256– 261.
Informatics In press. Cominesi, S. R., Farina, M., Giulioni, L., Picasso, B., & Scattolini, R. (2018). A two-layer
Bahramara, S., Moghaddam, M. P., & Haghifam, M. R. (2016). A bi-level optimization stochastic model predictive control scheme for microgrids. IEEE Transactions on
model for operation of distribution networks with micro-grids. Electrical Power and Control Systems Technology, 26 (1), 1– 13.
Energy Systems, 82, 169– 178. Cook, M. D., Parker, G. G., Robinett, R. D., & Weaver, W. W. (2017). Decentralized
Bella, A. L., Cominesi, S. R., Sandroni, C., & Scattolini, R. (2017). Hierarchical pre- mode-adaptive guidance and control for DC microgrid. IEEE Transactions on Power
dictive control of microgrids in islanded operation. IEEE Transactions on Automation Delivery, 32 (1), 263– 271.
Science and Engineering, 14 (2), 536– 546. Costabeber, A., Tenti, P., Erseghe, T., Tomasin, S., & Mattavelli, P. (2011). Distributed
Berkel, F., Görges, D., & Liu, S. (2013). Load-frequency control, economic dispatch and control of smart microgrids by dynamic grid mapping. Proceedings of the thirty seventh
unit commitment in smart microgrids based on hierarchical MPC. Proceedings of the annual conference of the IEEE industrial electronics society (pp. 1323–1328). .
IEEE conference on decision and control (pp. 2326–2333). . Cucuzzella, M., Incremona, G. P., & Ferrara, A. (2017). Decentralized sliding mode
Bersani, C., Dagdougui, H., Ouammi, A., & Sacile, R. (2017). Distributed robust control control of islanded ac microgrids with arbitrary topology. IEEE Transactions on
of the power flows in a team of cooperating microgrids. IEEE Transactions on Control Industrial Electronics, 64 (8), 6706– 6713.
Systems Technology, 25 (4), 1473– 1479. Deckmyn, C., Vandoorn, T. L., Vyver, J., Desmet, J., & Vandevelde, L. (2017). A
Bhattacharya, S., & Mishra, S. (2016). Coordinated decentralized control for PV-EV microgrid multilayer control concept for optimal power scheduling and voltage
based grid connected microgrids. Proceedings of the sixth IEEE international conference control. IEEE Transactions on Smart Grid In press.
on power systems (pp. 1–6). . Dehkordi, N. M., Sadati, N., & Hamzeh, M. (2017). Distributed robust finite-time sec-
Bhattarai, B. P., Bak-Jensen, B., Mahat, P., Pillai, J. R., & Maier, M. (2013). ondary voltage and frequency control of islanded microgrids. IEEE Transactions on
Hierarchical control architecture for demand response in smart grids. Proceedings of Power Systems In press.
the IEEE PES Asia-pacific power and energy engineering conference (pp. 1–6). . Deng, N., Zhang, X. P., Wang, P., Gu, X., & Wu, M. (2013). A converter-based general
Bidram, A., & Davoudi, A. (2012). Hierarchical structure of microgrids control system. interface for ac microgrid integrating to the grid. Proceedings of the fourth IEEE PES
IEEE Transactions on Smart Grid, 3 (4), 1963– 1976. innovative smart grid technologies (pp. 1–5). .
Bidram, A., Davoudi, A., Lewis, F. L., & Guerrero, J. M. (2013a). Distributed co- Diaz, N. L., Luna, A. C., Vasquez, J. C., & Guerrero, J. M. (2017a). Centralized control
operative secondary control of microgrids using feedback linearization. IEEE architecture for coordination of distributed renewable generation and energy storage
Transactions on Power Systems, 28 (3), 3462– 3470. in islanded ac microgrids. IEEE Transactions on Power Electronics, 32 (7), 5202–
Bidram, A., Davoudi, A., Lewis, F. L., & Ge, S. S. (2014a). Distributed adaptive voltage 5213.
control of inverter-based microgrids. IEEE Transactions on Energy Conversion, 29 (4), Diaz, N. L., Vasquez, J. C., & Guerrero, J. M. (2018). A communication less distributed
862– 872. control architecture for islanded microgrids with renewable generation and storage.
IEEE Transactions on Power Electronics, 33 (3), 1922– 1939.

147
S. Sen, V. Kumar Annual Reviews in Control 45 (2018) 118–151

Ding, M., Zhang, Y. Y., Mao, M. Q., Yang, W., & Liu, X. P. (2010). Operation opti- Guo, F., Wen, C., Mao, J., & Song, Y. D. (2015). Distributed secondary voltage and
mization for microgrids under centralized control. Proceedings of the second interna- frequency restoration control of droop-controlled inverter-based microgrids. IEEE
tional symposium on power electronics for distributed generation systems (pp. 984–987). . Transactions on Industrial Electronics, 62 (7), 4355– 4364.
Ding, M., Ma, K., Bi, R., Mao, M., & Chang, L. (2013). A hierarchical control scheme Hajar, K., Hably, A., Bacha, S., Elrafhi, A., & Obeid, Z. (2016a). Optimal centralized
based on multi-agent system for islanded multi-microgrids. Proceedings of the fourth control application on microgrids. Proceedings of the third IEEE international con-
IEEE international symposium on power electronics for distributed generation systems (pp. ference on renewable energies for developing countries (pp. 1–6). .
1–5). . Hamidia, R. J., Livania, H., Hosseinianb, S. H., & Gharehpetian, G. B. (2015).
Divshali, P., Alimardani, A., Hosseinian, S., & Abedi, M. (2012). Decentralized co- Distributed cooperative control system for smart microgrids. Electric Power Systems
operative control strategy of microsources for stabilizing autonomous VSC-based Research, 130, 241– 250.
microgrids. IEEE Trans. on Power Systems, 27 (4), 1949– 1959. Han, H., Hou, X., Yang, J., Wu, J., Su, M., & Guerrero, J. M. (2016a). Review of power
Dorfler, F., Simpson-Porco, J. W., & Bullo, F. (2016). Breaking the hierarchy: Distributed sharing control strategies for islanding operation of ac microgrids. IEEE Transactions
control and economic optimality in microgrids. IEEE Transactions on Control of on Smart Grid, 7 (1), 200– 215.
Network Systems, 3 (3), 241– 253. Hashempour, M., Savaghebi, M., Vasquez, J. C., & Guerrero, J. M. (2015). Hierarchical
Dragicevic, T., Lu, X., Vasquez, J. C., & Guerrero, J. M. (2016). DC microgrids -Part I: A control for voltage harmonics compensation in multi-area microgrids. Proceedings of
review of control strategies and stabilization techniques. IEEE Transactions on Power the IEEE tenth international symposium on diagnostics for electrical machines, power
Electronics, 31 (7), 4876– 4891. electronics and drives (pp. 415–420). .
Duan, J., Wang, C., Xu, H., Liu, W., Peng, J., & Jiang, H. (2018). Distributed control of Hatziargyriou, N. (2014). Microgrid: architecture and control. Wiley & Sons Publications,
inverter-interfaced microgrids with bounded transient line currents. IEEE IEEE Press ISBN: 9781118720684.
Transactions on Industrial Informatics In press. He, J., Li, Y., Liang, B., & Wang, C. (2017). Inverse power factor droop control for
Eid, B. M., Rahim, N. A., Selvaraj, J., & Khateb, A. (2016). “Control methods and decentralized power sharing in series-connected micro-converters based islanding
objectives for electronically coupled distributed energy resources in microgrids: A microgrids. IEEE Transactions on Industrial Electronics, 64 (9), 7444– 7454.
review. IEEE Systems Journal, 10 (2), 446– 458. Hirscha, A., Parag, Y., & Guerrero, J. M. (2018). Microgrids: A review of technologies,
El-Taweel, N., & Farag, H. (2017). Voltage regulation in islanded microgrids using key drivers, and outstanding issues. Renewable and Sustainable Energy Reviews, 90,
distributed constraint satisfaction. IEEE Transactions on Smart Grid In press. 402– 411.
Engels, J., Almasalma, H., & Deconinck, G. (2016). A distributed gossip-based voltage Hossain, M. J., Lu, J., Mahmud, M. A., & Aziz, T. (2014a). Advanced decentralized DER
control algorithm for peer-to-peer microgrids. Proceedings of the IEEE international control for islanded microgrids. Proceedings of the Australasian Universities power
conference on smart grid communications (pp. 370–375). . engineering conference (pp. 1–5). .
Etemadi, A. H., Davison, E. J., & Iravani, R. (2014). A generalized decentralized robust Hossain, M. J., Mahmud, M. A., Milano, F., Bacha, S., & Hably, A. (2016). Design of
control of islanded microgrids. IEEE Transactions on Power Systems, 29 (6), 3102– robust distributed control for interconnected microgrids. IEEE Transactions on Smart
3113. Grid, 7 (6), 2724– 2735.
Fan, Y., Hu, G., & Egerstedt, M. (2017). Distributed reactive power sharing control for Hosseinimehr, T., Ghosh, A., & Shahnia, F. (2016). Cooperative control of battery en-
microgrids with event-triggered communication. IEEE Transactions on Control Systems ergy storage systems in microgrids. Electrical Power and Energy Systems, 87, 109–
Technology, 25 (1), 118– 128. 120.
Farzin, H., Firuzabad, M., & Moeini-Aghtaie, M. (2016). Enhancing power system re- Hu, X., Zhou, H., Liu, Z. W., Guan, Z. H., & Chi, M. (2016a). Reactive power com-
silience through hierarchical outage management in multi-microgrids. IEEE pensation in microgrids via distributed control strategy. Proceedings of the twelfth
Transactions on Smart Grid, 7 (6), 2869– 2879. world congress on intelligent control and automation (pp. 1635–1640). .
Farzin, H., Firuzabad, M., & Moeini-Aghtaie, M. (2017a). Stochastic energy manage- Huang, P., Liu, P., Xiao, W., & Moursi, M. S. (2015). A novel droop-based average
ment of microgrids during unscheduled islanding period. IEEE Transactions on voltage sharing control strategy for dc microgrids. IEEE Transactions on Smart Grid, 6
Industrial Informatics, 13 (3), 1079– 1087. (3), 1096– 1106.
Farzin, H., Firuzabad, M., & Moeini-Aghtaie, M. (2017b). A stochastic multi-objective Iggland, E., Wiget, R., Chatzivasileiadis, S., & Anderson, G. (2015). Multi-area DC-OPF
framework for optimal scheduling of energy storage systems in microgrids. IEEE for HVAC and HVDC grids. IEEE Transactions on Power Systems, 30 (5), 2450– 2459.
Transactions on Smart Grid, 8 (1), 117– 127. Jana, K., Ray, A., Majoumerd, M. M., Assadi, M., & De, S. (2017). Polygeneration as a
Feng, W., Sun, K., Guan, Y., Guerrero, J. M., & Xiao, X. (2017). Active power quality future sustainable energy solution – a comprehensive review. Applied Energy, 201,
improvement strategy for grid-connected microgrid based on hierarchical control. 88– 111.
IEEE Transactions on Smart Grid In press. Jin, C., Wang, P., Xiao, J., Tang, Y., & Choo, F. H. (2014). Implementation of hier-
García-Villalobos, J., Zamora, I., Martín, J. I., Asensio, F. J., & Aperribay, V. (2014). archical control in dc microgrids. IEEE Transactions on Industrial Electronics, 61 (8),
Plug-in electric vehicles in electric distribution networks: A review of smart charging 4032– 4042.
approaches. Renewable and Sustainable Energy Reviews, 38, 717– 731. Kammer, C., & Karimi, A. (2016). Robust distributed averaging frequency control of
Gavriluta, C., Candela, J. I., Citro, C., Rocabert, J., Luna, A., & Rodríguez, P. (2014). inverter-based microgrids. Proceedings of the fifty fifth IEEE conference on decision and
Decentralized primary control of MTDC networks with energy storage and distributed control (pp. 4973–4978). .
generation. IEEE Transactions on Industry Applications, 50 (6), 4122– 4131. Karami, M., Seifi, H., & Mohammadian, M. (2016). Seamless control scheme for dis-
Gil, N., & Lopes, J. (2007). Hierarchical frequency control scheme for islanded multi- tributed energy resources in microgrids. IET Generation, Transmission & Distribution,
microgrids operation. Proceedings of the IEEE Lausanne power tech (pp. 473–478). . 10 (11), 2756– 2763.
Gkavanoudis, S. I., Oureilidis, K. O., & Demoulias, C. S. (2016). An adaptive droop Karavas, C. S., Kyriakarakos, G., Arvanitis, K. G., & Papadaki, G. (2015a). A multi-agent
control method for balancing the SoC of distributed batteries in AC microgrids. decentralized energy management system based on distributed intelligence for the
Proceedings of the IEEE seventeenth workshop on control and modeling for power elec- design and control of autonomous polygeneration microgrids. Energy Conversion and
tronics (pp. 1–6). . Management, 103, 166– 179.
Golsorkhi, M. S., & Lu, D. D. C. (2014). A decentralized power flow control method for Karimi, M., Wall, P., Mokhlis, H., & Terzija, V. (2017a). A new centralized adaptive
islanded microgrids using V-I droop. Proceedings of the twenty second Iranian con- underfrequency load shedding controller for microgrids based on a distribution state
ference on electrical engineering (pp. 604–609). . estimator. IEEE Transactions on Power Delivery, 32 (1), 370– 380.
Golsorkhi, M. S., & Lu, D. (2015). A control method for inverter-based islanded mi- Karimi, Y., Oraee, H., Golsorkhi, M. S., & Guerrero, J. (2017b). Decentralized method
crogrids based on V-I droop characteristics. IEEE Transactions on Power Delivery, 30 for load sharing and power management in a PV/battery hybrid source islanded
(3), 1196– 1204. microgrid. IEEE Transactions on Power Electronics, 32 (5), 3525– 3535.
Golsorkhi, M. S., & Lu, D. (2016). A decentralized control method for islanded micro- Katiraei, F., Iravani, M. R., & Lehn, P. W. (2005). Microgrid autonomous operation
grids under unbalanced conditions. IEEE Transactions on Power Delivery, 31 (3), during and subsequent to islanding process. IEEE Transaction on Power Delivery, 20
1112– 1121. (1), 248– 257.
Golsorkhi, M. S., Lu, D. D. C., Savaghebi, M., Vasquez, J. C., & Guerrero, J. M. (2016). A Kaur, A., Kaushal, J., & Basak, P. (2015). A review on microgrid central controller.
GPS-based control method for load sharing and power quality improvement in mi- Renewable and Sustainable Energy Reviews, 55, 338– 345.
crogrids. Proceedings of the IEEE eighth international power electronics and motion Khooban, M. H., Niknam, T., Shasadeghi, M., Dragicevic, T., & Blaabjerg, F. (2018).
control conference (pp. 3734–3740). . Load frequency control in microgrids based on a stochastic noninteger controller.
Golsorkhi, M. S., Lu, D., & Guerrero, J. M. (2017a). A GPS-based decentralized control IEEE Transactions on Sustainable Energy, 9 (2), 853– 861.
method for islanded microgrids. IEEE Transactions on Power Electronics, 32 (2), Khorsandi, A., Ashourloo, M., & Mokhtari, H. (2014). A decentralized control method
1615– 1625. for a low-voltage dc microgrid. IEEE Transactions on Energy Conversion, 29 (4), 793–
Green, R. C. , II, Wang, L., & Alam, M. (2011). The impact of plug-in hybrid electric 801.
vehicles on distribution networks: A review and outlook. Renewable and Sustainable Kou, P., Liang, D., & Gao, L. (2016). Distributed EMPC of multiple MGs for coordinated
Energy Reviews, 15, 544– 553. stochastic energy management. Applied Energy, 185, 939– 952.
Gu, W., Lou, G., Tan, W., & Yuan, X. (2017). A nonlinear state estimator-based de- Kou, P., Liang, D., & Gao, L. (2017). Distributed coordination of multiple PMSGs in an
centralized secondary voltage control scheme for autonomous microgrids. IEEE islanded dc microgrid for load sharing. IEEE Transactions on Energy Conversion, 32
Transactions on Power Systems, 32 (6), 4794– 4804. (2), 471– 485.
Guerrero, J. M., Matas, J., Vicuna, L. G., Castilla, M., & Miret, J. (2007). Decentralized Kyriakarakos, G., Piromalis, D. D., Dounis, A. I., Arvanitis, K. G., & Papadakis, G.
control for parallel operation of distributed generation inverters in microgrids using (2013). Intelligent demand side energy management system for autonomous poly-
resistive output impedance. IEEE Transactions on Industrial Electronics, 54 (2), 994– generation microgrids. Applied Energy, 103, 39– 51.
1004. Lai, J., Zhou, H., Lu, X., Yu, X., & Hu, W. (2016a). Droop-based distributed cooperative
Guerrero, J. M., Chandorkar, M., Lee, T., & Loh, P. C. (2013). Advanced control ar- control for microgrids with time-varying delays. IEEE Transactions on Smart Grid, 7
chitectures for intelligent microgrids—Part I: Decentralized and hierarchical control. (4), 1775– 1789.
IEEE Transactions on Industrial Electronics, 60 (4), 1254– 1262. Systems & control for the future of humanity, research agenda: Current and future roles,

148
S. Sen, V. Kumar Annual Reviews in Control 45 (2018) 118–151

impact and grand challenges. F. Lamnabhi-Lagarrigue (Ed.). Systems & control for Meng, L., Sanseverino, E. R., Luna, A., Dragicevic, T., Vasquez, J. C., & Guerrero, J. M.
the future of humanity, research agenda: Current and future roles, impact and grand (2016a). Microgrid supervisory controllers and energy management systems: A lit-
challenges, Annual Reviews in Control, 43, 1– 64. erature review. Renewable and Sustainable Energy Reviews, 60, 1263– 1273.
Lee, Z., Zhu, S., Zheng, J., Choi, D. H., & Wei, L. (2014). Research on dynamic process Meng, W., Wang, X., & Liu, S. (2018). Distributed load sharing of an inverter-based
of dc microgrid under hierarchical control. Proceedings of the nineteenth world congress microgrid with reduced communication. IEEE Transactions on Smart Grid, 9 (2),
the international federation of automatic control (pp. 4560–4564). . 1354– 1364.
Li, C., Dragicevic, T., Plaza, M., Andrade, F., Vasquez, J., & Guerrero, J. (2014). Moayedi, S., & Davoudi, A. (2017). Unifying distributed dynamic optimization and
Multiagent based distributed control for state-of-charge balance of distributed energy control of islanded dc microgrids. IEEE Transactions on Power Electronics, 32 (3),
storage in dc microgrids. Proceedings of the fortieth annual conference of the IEEE in- 2329– 2346.
dustrial electronics society (pp. 2180–2184). . Mojica-Nava, E., Rivera, S., & Quijano, N. (2016). Distributed dispatch control in mi-
Li, Y., Liu, N., & Zhang, J. (2015). Jointly optimization and distributed control for crogrids with network losses. Proceedings of the IEEE Conference on Conference
interconnected operation of autonomous microgrids. Proceedings of the IEEE in- Application (pp. 285–290). .
novative smart grid technologies (pp. 1–6). . Moradi, M. H., Eskandari, M., & Hosseinian, S. M. (2015). Cooperative control strategy
Li, Q., Chen, F., Chen, M., Guerrero, J. M., & Abbott, D. (2016a). Agent-based de- of energy storage systems and micro sources for stabilizing microgrids in different
centralized control method for islanded microgrids. IEEE Transactions on Smart Grid, operation modes. Electrical Power and Energy Systems, 78, 390– 400.
7 (2), 637– 649. Morais, H., Sousa, T., Vale, Z., & Faria, P. (2014). Evaluation of the electric vehicle
Li, Q., Peng, C., Chen, M., Chen, F., Kang, W., Guerrero, J., et al. (2017a). Networked impact in the power demand curve in a smart grid environment. Energy Conversion
and distributed control method with optimal power dispatch for islanded microgrids. and Management, 82, 268– 282.
IEEE Transactions on Industrial Electronics, 64 (1), 493– 504. Morstyn, T., Savkin, A. V., Hredzak, B., & Agelidis, V. G. (2017). Multi-agent sliding
Li, Q., Peng, C., Wang, M., Chen, M., Guerrero, J. M., & Abbott, D. (2018a). Distributed mode control for state of charge balancing between battery energy storage systems
secondary control and management of islanded microgrids via dynamic weights. distributed in a dc microgrid. IEEE Transactions on Smart Grid In press.
IEEE Transactions on Smart Grid In press. Mousavi, S. Y. M., Jalilian, A., Savaghebi, M., & Guerrero, J. M. (2018). Autonomous
Lidula, N. W. A., & Rajapakse, A. D. (2011). Microgrids research: A review of experi- control of current and voltage controlled dg interface inverters for reactive power
mental microgrids and test systems. Renewable and Sustainable Energy Reviews, 15, sharing and harmonics compensation in islanded microgrids. IEEE Transactions on
186– 202. Power Electronics In press.
Lin, P., Jin, C., Xiao, J., Li, X., Shi, D., Tang, Y., et al. (2018). A Distributed control Nakazawa, C., Murakami, K., Shimasaki, Y., Hayashi, N., & Nakanishi, Y. (2013).
architecture for global system economic operation in autonomous hybrid ac/dc mi- Hierarchical operation and control for microgrids. Proceedings of the twenty second
crogrids. IEEE Transactions on Smart Grid In press. international conference on electricity distribution (pp. 1–4). .
Liu, W., Gu, W., Sheng, W., Meng, X., Wu, Z., & Chen, W. (2014). Decentralized multi- Nasirian, V., Davoudi, A., & Lewis, F. L. (2014). Distributed adaptive droop control for
agent system-based cooperative frequency control for autonomous microgrids with DC microgrids. Proceedings of the IEEE applied power electronics conference and ex-
communication constraints. IEEE Transactions on Sustainable Energy, 5 (2), 446– position (pp. 1147–1152). .
456. Nasirian, V., Moayedi, S., Davoudi, A., & Lewis, F. L. (2015). Distributed cooperative
Liu, W., Gu, W., Xu, Y., Xue, S., Chen, M., Zhao, B., et al. (2015). Improved average control of dc microgrids. IEEE Transactions on Power Electronics, 30 (4), 2288– 2303.
consensus algorithm based distributed cost optimization for loading shedding of Nasirian, V., Shafiee, Q., Guerrero, J. M., Lewis, F. L., & Davoudi, A. (2016). Droop-
autonomous microgrids. Electrical Power and Energy Systems, 73, 89– 96. Free Distributed Control for AC Microgrids. IEEE Transactions on Power Electronics,
Liu, W., Gu, W., Sheng, W., Meng, X., Xue, S., & Chen, M. (2016). Pinning-based 31 (2), 1600– 1617.
distributed cooperative control for autonomous microgrids under uncertain com- Nasirian, V., Yadav, A. P., Lewis, F. L., & Davoudi, A. (2017). Distributed assistive
munication topologies. IEEE Transactions on Power Systems, 31 (2), 1320– 1329. control of power buffers in dc microgrids. IEEE Transactions on Energy Conversion, 32
Lizhen, W., Xiaohong, H., Zhengzhe, L., Zuohao, L., & Hu, Z. (2016). Decentralized (4), 1396– 1406.
control approach for voltage unbalance compensation in islanded microgrid. Nosratabadi, S. M., Hooshmand, R. A., & Gholipour, E. (2017). A comprehensive review
Proceedings of the IEEE eighth international power electronics and motion control con- on microgrid and virtual power plant concepts employed for distributed energy re-
ference (pp. 1776–1781). . sources scheduling in power systems. Renewable and Sustainable Energy Reviews, 67,
Lorzadeh, I., Savaghebi, M., Abyaneh, H., & Guerrero, J. (2015). A hierarchical control 341– 363.
scheme for reactive power and harmonic current sharing in islanded microgrids. Nudell, T. R., Brignone, M., Robba, M., Bonfiglio, A., Delfino, F., & Annaswamy, A.
Proceedings of the seventeenth European conference on power electronics and applications (2017). A dynamic market mechanism for combined heat and power microgrid
(pp. 1–10). . energy management. Proceedings of IFAC World Congress 50. Proceedings of IFAC
Lou, G., Gu, W., Xu, Y., Cheng, M., & Liu, W. (2017). Distributed MPC-based secondary World Congress (pp. 10033–10039). .
voltage control scheme for autonomous droop-controlled microgrids. IEEE Nutkani, I.,U., Loh, P. C., & Blaabjerg, F. (2012). Distributed control of hybrid ac
Transactions on Sustainable Energy, 8 (2), 792– 804. microgrids with dynamic active and reactive power capacity tuning. Proceedings of
Lu, X., Guerrero, J. M., Sun, K., Vasquez, J. C., Teodorescu, R., & Huang, L. (2014). the IEEE energy conversion congress and exposition (pp. 1954–1959). .
Hierarchical control of parallel ac-dc converter interfaces for hybrid microgrids. IEEE Nutkani, I.,U., Loh, P. C., Wang, P., & Blaabjerg, F. (2017). Decentralized economic
Transactions on Smart Grid, 5 (2), 683– 692. dispatch scheme with online power reserve for microgrids. IEEE Transactions on
Lu, L., Liu, H. J., & Zhu, H. (2016a). Distributed secondary control for isolated mi- Smart Grid, 8 (1), 139– 148.
crogrids under malicious attacks. Proceedings of the North American power symposium Olivares, D. E., Cañizares, C. A., & Kazerani, M. (2011). A centralized optimal energy
(pp. 1–6). . management system for microgrids. Proceedings of the IEEE PES General Meeting (pp.
Lu, X., Yu, X., Lai, J., Guerrero, J. M., & Zhou, H. (2017a). Distributed secondary 1–6). .
voltage and frequency control for islanded microgrids with uncertain communication Olivares, D. E., Cañizares, C. A., & Kazerani, M. (2014). A centralized energy man-
links. IEEE Transactions on Industrial Informatics, 13 (2), 448– 460. agement system for isolated microgrids. IEEE Transactions on Smart Grid, 5 (4),
Lu, T., Wang, Z., Ai, Q., & Lee, W. J. (2017b). Interactive model for energy management 1864– 1875.
of clustered microgrids. IEEE Transactions on Industry Applications, 53 (3), 1739– Oliveira, T. R., Silva, W. W., & Donoso-Garcia, P. F. (2017). Distributed secondary level
1750. control for energy storage management in dc microgrids. IEEE Transactions on Smart
Luo, F., Lai, Y. M., Loo, K. H., Tse, C. K., & Ruan, X. (2013). A generalized droop- Grid, 8 (6), 2591– 2607.
control scheme for decentralized control of inverter-interfaced microgrids. Oureilidis, K. O., & Demoulias, C. S. (2015). A decentralized impedance-based adaptive
Proceedings of the IEEE international symposium on circuits and systems (pp. 1320– droop method for power loss reduction in a converter-dominated islanded microgrid.
1323). . Sustainable Energy, Grids and Networks, 5, 39– 49.
Lv, T., Ai, Q., & Zhao, Y. (2015). A bi-level multi-objective optimal operation of grid- Ovalle, A., Ramos, G., Bacha, S., Hably, A., & Rumeau, A. (2015). Decentralized control
connected microgrids. Electric Power System Research, 131, 60– 70. of voltage source converters in microgrids based on the application of instantaneous
Ma, J., Yuan, L., Zhao, Z., & He, F. (2017). Transmission loss optimization-based power theory. IEEE Transactions on Industrial Electronics, 62 (2), 1152– 1162.
optimal power flow strategy by hierarchical control for dc microgrids. IEEE Palizban, O., Kauhaniemi, K., & Guerrero, J. M. (2014). Microgrids in active network
Transactions on Power Electronics, 32 (3), 1952– 1963. management—Part I: Hierarchical control, energy storage, virtual power plants, and
Majumder, R. (2010). Modeling, stability analysis and control of microgrid Ph.D thesis market participation. Renewable and Sustainable Energy Reviews, 36, 428– 439.
Australia: Faculty of Build and Environment Engineering., Queensland University of Parhizi, S., Khodaei, A., & Shahidehpour, M. (2018). Market-based versus price-based
Technology. microgrid optimal scheduling. IEEE Transactions on Smart Grid, 9 (2), 615– 623.
Maknouninejad, A., Qu, Z., Lewis, F. L., & Davoudi, A. (2014). Optimal, nonlinear, and Peyghami, S., Mokhtari, H., & Blaabjerg, F. (2017). Decentralized load sharing in a low
distributed designs of droop controls for dc microgrids. IEEE Transactions on Smart voltage direct current microgrid with an adaptive droop approach based on a su-
Grid, 5 (5), 2508– 2516. perimposed frequency. IEEE Journal of Emerging and Selected Topics in Power
Malik, S. M., Ai, X., Sun, Y., Zhengqi, C., & Shupeng, Z. (2016). Voltage and frequency Electronics, 5 (3), 1205– 1215.
control strategies of hybrid ac/dc microgrid: A review. IET Generation, Transmission & Peyghami-Akhuleh, S., Mokhtari, H., Loh, P. C., & Blaabjerg, F. (2016). Distributed
Distribution, 11 (2), 303– 313. secondary control in dc microgrids with low-bandwidth communication link.
Martin-Martinez, F., Sánchez-Miralles, A., & Rivier, M. (2016). A literature review of Proceedings of the seventh power electronics and drive system technology conference (pp.
microgrids: A functional layer based classification. Renewable and Sustainable Energy 641–645). .
Reviews, 62, 1133– 1153. Pilloni, A., Pisano, A., & Usai, E. (2018). Robust finite-time frequency and voltage
Meng, L., Dragicevic, T., Vasquez, J. C., Guerrero, J. M., & Sanseverino, E. R. (2015). restoration of inverter-based microgrids via sliding-mode cooperative control. IEEE
Hierarchical control with virtual resistance optimization for efficiency enhancement Transactions on Industrial Electronics, 65 (1), 907– 917.
and state of charge balancing in dc microgrids. Proceedings of the IEEE first interna- Qian, H., Zhang, J., Lai, J., & Yu, W. (2011). A high-efficiency grid-tie battery energy
tional conference on DC microgrids (pp. 1–6). . storage system. IEEE Trans. on Power Electronics, 26 (3), 886– 896.

149
S. Sen, V. Kumar Annual Reviews in Control 45 (2018) 118–151

Rahman, M. S., Mahmud, M. A., Oo, A. M. T., & Orchi, T. F. (2015). Distributed agent- improvement using micro grid voltage controller. Proceedings of the IEEE southeastcon
based control scheme for single-phase parallel inverters in microgrids with photo- (pp. 1–6). .
voltaic systems. Proceedings of the Australasian Universities power engineering con- Tan, K. T., Peng, X. Y., So, P. L., Chu, Y. C., & Chun, M. Z. Q. (2012). Centralized
ference (pp. 1–6). . control for parallel operation of distributed generation inverters in microgrids. IEEE
Ramezani, M., & Li, S. (2016). Voltage and frequency control of islanded microgrid Transactions on Smart Grid, 3 (4), 1977– 1987.
based on combined direct current vector control and droop control. Proceedings of the Tan, K., So, P., Chu, Y., & Chen, M. (2013). A flexible ac distribution system device for a
IEEE PES general meeting (pp. 1–5). . microgrid. IEEE Transactions on Energy Conversion, 28 (3), 601– 610.
Rezaei, N., & Kalantar, M. (2014). Economic–environmental hierarchical frequency Tavakoli, A., Negnevitsky, M., & Muttaqi, K. M. (2017). A decentralized MPC for
management of a droop-controlled islanded microgrid. Energy Conversion and multiple distributed generators in islanded mode of operation. IEEE Transactions on
Management, 88, 498– 515. Industry Applications, 53 (2), 1466– 1575.
Rezaei, N., Ahmadi, A., Khazali, A., & Guerrero, J. M. (2018). Energy and frequency Tegling, E., Andreasson, M., Simpson-Porco, J. W., & Sandberg, H. (2016). Proceedings
hierarchical management system using information gap decision theory for islanded of the American control conference improving performance of droop-controlled microgrids
microgrids. IEEE Transactions on Industrial Electronics In press. through distributed PI-control (pp. 2321–2327). .
Riverso, S., Sarzo, F., & Ferrari-Trecate, G. (2015). Plug-and-play voltage and frequency Toub, M., Aniba, G., Maaroufi, M., & Robinet, R. D. , III (2015). Decentralized
control of islanded microgrids with meshed topology. IEEE Transactions on Smart Hamiltonian control of isolated ac microgrids: Theory & design. IEEE Innovative
Grid, 6 (3), 1176– 1184. Smart Grid Technologies - Asia (ISGT ASIA), 1– 6.
Sadabadi, M. S., Karimi, A., & Karimi, H. (2015). Fixed-order decentralized/ distributed Trung, T. D. (2013). A robust decentralized controller design for inverter-based
control of islanded inverter-interfaced microgrids. Control Engineering Practice, 45, Microgrids with dynamic loads. Proceedings of the annual conference of the IEEE in-
174– 193. dustrial electronics society (pp. 1507–1512). .
Sadabadi, M. S., Shafiee, Q., & Karimi, A. (2017). Plug-and-play voltage stabilization in Tsikalakis, A., & Hatziargyriou, N. (2008). Centralized control for optimizing microgrids
inverter-interfaced microgrids via a robust control strategy. IEEE Transactions on operation. IEEE Transactions on Energy Conversion, 23 (1), 241– 248.
Control Systems Technology, 25 (3), 781– 791. Tucci, M., Riverso, S., Vasquez, J. C., Guerrero, J. M., & Ferrari-Trecate, G. (2016). A
Sahoo, S., Mishra, S., & Padhy, N. P. (2016). A decentralized adaptive droop based decentralized scalable approach to voltage control of dc islanded microgrids. IEEE
power management scheme in autonomous DC microgrid. Proceedings of the IEEE PES Transactions on Control Systems Technology, 24 (6), 1965– 1979.
Asia-pacific power and energy conference (pp. 1018–1022). . Ustun, T. S., Ozansoy, C., & Zayegh, A. (2011). Recent developments in microgrids and
Sakurama, K., & Miura, M. (2017). Communication-based decentralized demand re- example cases around the world-A review. Renewable and Sustainable Energy Reviews,
sponse for smart microgrids. IEEE Transactions on Industrial Electronics, 64 (6), 15, 4030– 4041.
5192– 5202. Vu, T. V., Perkinsa, D., Diaza, F., Gonsoulina, D., Edringtona, C. S., & El-Mezyani, T.
Sanseverino, E., Silvestre, M., Quang, N., Mineo, L., Guerrero, J., Luna, A., et al. (2017a). Robust adaptive droop control for DC microgrids”. Electric Power Systems
(2015). Energy management systems and tertiary regulation in hierarchical control Research, 146, 95– 106.
architectures for islanded microgrids. Proceedings of the IEEE fifteenth international Wang, X., Blaabjerg, F., Chen, Z., & Guerrero, J. (2011). A centralized control archi-
conference on environment and electrical engineering (pp. 144–149). . tecture for harmonic voltage suppression in islanded microgrids. Proceedings of the
Savaghebi, M., Guerrero, J. M., Jalilian, A., & Vasquez, J. C. (2011). Hierarchical thirty seventh annual conference of the IEEE industrial electronics society (pp. 3070–
control scheme for voltage unbalance compensation in islanded microgrids. 3075). .
Proceedings of the thirty seventh annual conference of the IEEE industrial electronics so- Wang, P., Xiao, J., Setyawan, L., Jin, C., & Hoong, C. F. (2014). Hierarchical control of
ciety (pp. 3158–3163). . active hybrid energy storage system in dc microgrids. Proceedings of the ninth IEEE
Savaghebi, M., Hashempour, M., & Guerrero, J. M. (2014). Hierarchical coordinated conference on industrial electronics and applications (pp. 569–574). .
control of distributed generators and active power filters to enhance power quality of Wang, Y., Chen, Z., Wang, X., Tian, Y., Tan, Y., & Yang, C. (2015a). An estimator-based
microgrids. Proceedings of the fifty fifth international scientific conference on power and distributed voltage-predictive control strategy for ac islanded microgrids. IEEE
electrical engineering of Riga technical university (pp. 259–264). . Transactions on Power Electronics, 30 (7), 3934– 3951.
Schiffer, J., Seel, T., Raisch, J., & Sezi, T. (2014). A consensus-based distributed voltage Wang, Z., Wu, W., & Zhang, B. (2015b). A fully distributed active power control method
control for reactive power sharing in microgrids. Proceedings of the European control with minimum generation cost in grid-connected microgrids. Proceedings of the IEEE
conference (pp. 1299–1305). . PES general meeting (pp. 1–5). .
Schiffer, J., Seel, T., Raisch, J., & Sezi, T. (2016). Voltage stability and reactive power Wang, Z., Sun, H., & Nikovski, D. (2016a). Distributed secondary voltage controller for
sharing in inverter-based microgrids with consensus-based distributed voltage con- droop-controlled microgrids to improve power quality in power distribution net-
trol. IEEE Transactions on Control Systems Technology, 24 (1), 96– 109. works. Proceedings of the IEEE PES general meeting (pp. 1–5). .
Serban, I. (2014). Frequency restoration in microgrids by means of distributed control Wang, Z., Wu, W., & Zhang, B. (2016c). A distributed control method with minimum
with minimum communication requirements. Proceedings of the IEEE twenty third generation cost for dc microgrids. IEEE Transactions on Energy Conversion, 31 (4),
international symposium on industrial electronics (pp. 2590–2595). . 1462– 1470.
Shafie-khah, M., Neyestani, N., Damavandi, M. Y., Gil, F. A. S., & Catalão, J. P. S. Wang, Y., Wang, X., Chen, Z., & Blaabjerg, F. (2017a). Distributed optimal control of
(2016). Economic and technical aspects of plug-in electric vehicles in electricity reactive power and voltage in islanded microgrids. IEEE Transactions on Industry
markets. Renewable and Sustainable Energy Reviews, 53, 1168– 1177. Applications, 53 (1), 340– 349.
Shafiee, Q., Vasquez, J. C., & Guerrero, J. M. (2012). Distributed secondary control for Wang, Z., Wu, W., & Zhang, B. (2017b). A distributed quasi-newton method for droop-
islanded microgrids - a networked control systems approach. Proceedings of the thirty free primary frequency control in autonomous microgrids. IEEE Transactions on Smart
eighth annual conference on IEEE industrial electronics society (pp. 5637–5642). . Grid In Press.
Shafiee, Q., Stefanovic, C., Dragicevic, T., Popovski, P., Vasquez, J. C., & Guerrero, J. Weaver, W. W., Robinett, R. D. , III, Parker, G. G., & Wilson, D. G. (2015). Distributed
M. (2014a). Robust networked control scheme for distributed secondary control of control and energy storage requirements of networked dc microgrids. Control
islanded microgrids. IEEE Transactions on Industrial Electronics, 61 (10), 5363– 5374. Engineering Practice, 44, 10– 19.
Shafiee, Q., Vasquez, J. C., Guerrero, J. M., Nasirian, V., & Davoudi, A. (2015). Wu, X., & Shen, C. (2017). Distributed optimal control for stability enhancement of
Cooperative frequency control for autonomous ac microgrids. Proceedings of the IEEE microgrids with multiple distributed generators. IEEE Transactions on Power Systems,
Eindhoven Power Technology (pp. 1–6). . 32 (5), 4045– 4059.
Shafiee, Q., Nasirian, V., Vasquez, J. C., Guerrero, J. M., & Davoudi, A. (2017). A multi- Wu, D., Tang, F., Dragicevic, T., Vasquez, J. C., & Guerrero, J. M. (2013). Coordinated
functional fully distributed control framework for ac microgrids. IEEE Transactions on primary and secondary control with frequency-bus-signaling for distributed genera-
Smart Grid In press. tion and storage in islanded microgrids. Proceedings of the thirty ninth annual con-
Simpson-Porco, J. W., Shafiee, Q., Bullo, F., Shafiee, Q., & Guerrero, J. M. (2013). ference of the IEEE industrial electronics society (pp. 7140–7145). .
Stability, power sharing, & distributed secondary control in droop-controlled mi- Wu, D., Tang, F., Dragicevic, T., Vasquez, J., & Guerrero, J. (2015). A control ar-
crogrids. Proceedings of the IEEE international conference on smart grid communications chitecture to coordinate renewable energy sources and energy storage systems in
(pp. 672–677). . islanded microgrids. IEEE Transactions on Smart Grid, 6 (3), 1156– 1166.
Simpson-Porco, J. W., Shafiee, Q., Dorfler, F., Vasquez, J. C., Guerrero, J. M., & Bullo, Wu, X., Shen, C., & Iravani, R. (2017). A distributed, cooperative frequency and voltage
F. (2015). Secondary frequency and voltage control of islanded microgrids via dis- control for microgrids. IEEE Transactions on Smart Grid In press.
tributed averaging. IEEE Transactions on Industrial Electronics, 62 (11), 7025– 7038. Xia, Y., Peng, Y., Hu, H., Wang, Y., & Wei, W. (2016). Advanced unified decentralised
Soshinskaya, M., Graus, W., Guerrero, J., & Vasquez, J. (2014). Microgrids: control method with voltage restoration for DC microgrids. IET Renewable Power
Experiences, barriers and success factors. Renewable and Sustainable Energy Reviews, Generation, 10 (6), 861– 871.
40, 659– 672. Xia, Y., Wei, W., Yu, M., Peng, Y., & Tang, J. (2018). Decentralized multi-time scale
Srinivasan, M., & Kwasinski, A. (2015). Autonomous hierarchical control of dc micro- power control for a hybrid ac/dc microgrid with multiple subgrids. IEEE Transactions
grids with constant-power loads. Proceedings of the IEEE applied power electronics on Power Electronics, 33 (5), 4061– 4072.
conference and exposition (pp. 2808–2815). . Xiao, J., Wang, P., & Setyawan, L. (2015). Hierarchical control of hybrid energy storage
Stadler, P., Ashouri, A., & Marechal, F. (2016). Distributed model predictive control of system in dc microgrids. IEEE Transactions on Industrial Electronics, 62 (8), 4915–
energy systems in microgrids. Proceedings of the annual IEEE system conference (pp. 1– 4924.
6). . Xiao, J., Zhang, L., Wang, P., & Xu, Q. (2016). Distributed coordination for economical
Sun, Y., Zhong, C., Hou, X., Yang, J., Han, H., & Guerrero, J. M. (2016). Distributed operation of power park with multiple microgrids. Proceedings of the Asian conference
cooperative synchronization strategy for multi-bus microgrids. Electrical Power and on energy, power and transportation electrification (pp. 1–6). .
Energy Systems, 86, 18– 28. Xin, H., Zhang, L., Wang, Z., Gan, D., & Wong, K. P. (2015). Control of island ac
Sun, Y., Shi, G., Li, X., Yuan, W., Su, M., Han, H., et al. (2018). An f-P/Q droop control microgrids using a fully distributed approach. IEEE Transactions on Smart Grid, 6 (2),
in cascaded-type microgrid. IEEE Transactions on Power Systems, 33 (1), 1136– 1138. 943– 945.
Tabrizi, M. A., Radman, G., & Tamersi, A. (2012). Microgrid voltage profile Xin, H., Zhao, R., Zhang, L., Wang, Z., Wong, K. P., & Wei, W. (2016). A decentralized

150
S. Sen, V. Kumar Annual Reviews in Control 45 (2018) 118–151

hierarchical control structure and self-optimizing control strategy for f-P type DGs in Zaery, M., Ahmed, E. M., Orabi, M., & Abdelghani, A. M. (2015). Distributed cooperative
islanded microgrids. IEEE Transactions on Smart Grid, 7 (1), 3– 5. control with lower generation cost for dc microgrid. First Workshop on Smart Grid and
Xu, J., Zou, Y., & Niu, Y. (2013). Distributed predictive control for energy management Renewable Energy (pp. 1–6). .
of multi-microgrids systems. Proceedings of the thirteenth IFAC symposium on large Zaidi, A., & Kupzog, F. (2008). Microgrid automation - A self-configuring approach.
scale complex systems: theory and applications (pp. 551–556). . Proceedings of the IEEE international multitopic conference (pp. 565–570). .
Xu, X., Jia, H., Wang, D., Yu, D. C., & Chiang, H. (2015). Hierarchical energy man- Zamora, R., & Srivastava, A. K. (2018). Multi-layer architecture for voltage and fre-
agement system for multi-source multi-product microgrids. Renewable Energy, 78, quency control in networked microgrids. IEEE Transactions on Smart Grid, 9 (3),
621– 630. 2076– 2085.
Xu, Q., Wang, P., Zhang, Y., Wen, C., & Xiao, J. (2016a). A decentralized control Zhang, R., & Hredzak, B. (2018). Distributed finite-time multi-agent control for dc mi-
strategy for economic operation of autonomous ac microgrids. Proceedings of the forty crogrids with time delays. IEEE Trans. on Smart Grid In Press.
second annual conference of IEEE industrial electronics society (pp. 4020–4024). . Zhang, Z., Cai, Y., He, X., & Liu, Y. (2014). A distributed architecture based on micro-
Xu, Q., Xiao, J., Hu, X., Wang, P., & Lee, M. Y. (2017). A decentralized power man- bank modules with self-reconfiguration control regarding battery recovery effect for
agement strategy for hybrid energy storage system with autonomous bus voltage dc microgrids. Proceedings of the IEEE applied power electronics conference and ex-
restoration and state of charge recovery. IEEE Transactions on Industrial Electronics, position (pp. 3091–3096). .
64 (9), 7098– 7108. Zhang, Q., Liu, Y., Zhao, Y., & Wang, N. (2015). A multi-mode operation control
Xu, Y., Sun, H., Gu, W., Xu, Y., & Li, Z. (2018). Optimal distributed control for strategy for flexible microgrid based on sliding-mode direct voltage and hierarchical
secondary frequency and voltage regulation in an islanded microgrid. IEEE controls. ISA Transactions, 61, 188– 198.
Transactions on Industrial Informatics In press. Zhang, L., Xin, H., Wang, Z., & Gan, D. (2016). A decentralized quasi-hierarchical
Yang, J., Jin, X., Wu, X., Acuna, P., Aguilera, R. P., Morstyn, T., et al. (2016a). control scheme for droop controlled AC island microgrids. Proceedings of the IEEE
Decentralized control method for DC microgrids with improved current sharing ac- power and energy society general meeting (pp. 1–5). .
curacy. IET Generation, Transmission & Distribution, 11 (3), 696– 706. Zhang, H., Kim, S., Sun, Q., & Zhou, J. (2017a). Distributed adaptive virtual impedance
Yao, G., Lu, Y., Tang, T., Benbouzid, M., Zheng, Y., & Wang, T. (2013). A central control for accurate reactive power sharing based on consensus control in microgrids.
control strategy of parallel inverters in ac Microgrid. Proceedings of the thirty ninth IEEE Transaction on Smart Grid, 8 (4), 1749– 1761.
annual conference of the IEEE industrial electronics society (pp. 7112–7117). . Zhao, X., Savaghebi, M., Guerrero, J., Vasquez, J., Sun, K., Wu, X., et al. (2015).
Yazdanian, M., & Mehrizi-Sani, A. (2014). Distributed control techniques in microgrids. Negative sequence droop method based hierarchical control for low voltage ride-
IEEE Transactions on Smart Grid, 5 (6), 2901– 2909. through in grid-interactive microgrids. Proceedings of the IEEE energy conversion
Yoldas, Y., Onen, A., Muyeen, S., Vasilakos, A., & Alan, I. (2017). Enhancing smart grid congress and exposition (pp. 6896–6903). .
with microgrids: Challenges and opportunities. Renewable and Sustainable Energy Zhaoa, J., & Dorfler, F. (2015). Distributed control and optimization in dc microgrids.
Reviews, 72, 205– 214. Automatica, 61, 18– 26.
Yue, W., Zhao, C., Lu, Y., & Li, G. (2010). A scheme of connecting microgrid to ac grid Zholbaryssov, M., & Dominguez-Garcia, A. D. (2017). Distributed enforcement of phase-
via flexible power electronics interface. Proceedings of the international conference on cohesiveness for frequency control of islanded inverter-based microgrids. IEEE
power system technology (pp. 1–6). . Transactions on Control of Network Systems In press.
Zaery, M., Ahmed, E. M., & Orabi, M. (2016). Consensus algorithm based distributed Zhou, J., Kim, S., Zhang, H., Sun, Q., & Han, R. (2017). Consensus-based distributed
control for economic operation of islanded dc microgrids. Proceedings of the eighteenth control for accurate reactive, harmonic and imbalance power sharing in microgrids.
international middle east power systems conference (pp. 854–859). . IEEE Transactions on Smart Grid In press.

151

You might also like