energies

Article
An Improved Current Control Strategy for a
Grid-Connected Inverter under Distorted
Grid Conditions
Ngoc Bao Lai and Kyeong-Hwa Kim *
Department of Electrical and Information Engineering, Seoul National University of Science and Technology,
232 Gongneung-ro, Nowon-gu, Seoul 139-743, Korea; laingocbao@gmail.com
* Correspondence: k2h1@seoultech.ac.kr; Tel.: +82-2-970-6406; Fax: +82-2-978-2754

Academic Editor: Shuhui Li

Received: 13 November 2015; Accepted: 7 March 2016; Published: 12 March 2016

Abstract: This paper presents an improved current control strategy for a three-phase grid-connected
inverter under distorted grid conditions. The main challenge associated with the grid-connected
inverter in distributed generation (DG) systems is to maintain the harmonic contents in output
current below the specified values even when the grid is subject to uncertain disturbances such as
harmonic distortion. To overcome such a challenge, an improved current control scheme is proposed
for a grid-connected inverter, in which the fundamental and harmonic currents are independently
controlled by a proportional-integral (PI) decoupling controller and a predictive basis controller,
respectively. The controller design approach is based on the model decomposition method, where
the measured inverter currents and grid voltages are divided into the fundamental and harmonic
components by means of moving average filters (MAFs). Moreover, to detect the angular displacement
and angular frequency with better accuracy, even in the presence of the grid disturbance, the MAF is
also introduced to implement an enhanced phase-lock loop (PLL) structure. Theoretical analyses as
well as comparative simulation results demonstrate that the proposed control scheme can effectively
compensate the uncertainties caused by the grid voltages with fast transient response. To validate
the feasibility of the proposed scheme, the whole control algorithms are implemented on 2 kVA
three-phase grid-connected inverter system using 32-bit floating-point DSP TMS320F28335. As a
result, the proposed scheme is an attractive way to control a grid-connected inverter under adverse
grid conditions.

Keywords: current control; distorted grid conditions; DSP TMS320F28335; grid-connected inverter;
model decomposition; moving average filter

1. Introduction
Three-phase grid-connected inverters have been widely employed in various applications,
including renewable power generation and regenerative energy systems. This is due to the recent
development trend constructing the electrical grid in terms of distributed generation (DG) systems,
in which grid-connected inverters are connected in parallel with each other to form a microgrid [1,2].
The essential functionality of microgrids is to have the ability to operate either in grid-connected or
autonomous mode in case of the absence of the main grid. Another requirement is to handle effectively
the exchange of active and reactive power between the microgrid and main grid [3,4]. In order to
fulfill the aforementioned criteria, the control strategy of a microgrid is generally designed based on
three levels of hierarchical structure to provide smartness and flexibility to microgrid, which includes
the primary control, the secondary control, and the tertiary control [5]. To follow the development
trend of the microgrid, the grid-connected inverter should be able to not only provide stable operation

Energies 2016, 9, 190; doi:10.3390/en9030190 www.mdpi.com/journal/energies

Energies 2016, 9, 190 2 of 23

but also improve the system performance even during the presence of abnormal grid conditions.
In general, the current control loop, which belongs to the primary control, is responsible for the quality
of the exchanging power. According to the recently published utility interface standard regarding DG
systems [6,7], the harmonic contents in the injected current need to meet certain current-distortion levels
in order to guarantee the power quality as well as fulfill interconnection requirements. Consequently,
the current control loop should be designed to ensure the effective operation of the grid-connected
inverter under any operating conditions of grid.
The PI controller has been widely used as current controller in variety of inverter applications
because of its simplicity and stability. The design and implementation of this controller are quite
straightforward in continuous as well as in discrete-time domain [8]. Furthermore, low computational
burden of this control algorithm makes it easier to implement the whole control system with only one
digital signal controller. However, despite these advantages, the PI controllers have the limitations
that they are unable to cope with sinusoidal reference signals and periodic disturbances [9]. Although
the sinusoidal tracking problem can be completely solved by implementing the controller in the
synchronously rotating reference frame, poor disturbance rejection capability makes the PI type
controller unsuitable for current control strategy of a grid-connected inverter in the presence of the
distorted grid voltages.
For the purpose of eradicating the harmonics from inverter currents, several control approaches
have been proposed, which is categorized as selective and non-selective methods [10]. Proportional
resonant (PR) controller is widely used as selective harmonic compensation scheme [11]. This controller
is often implemented in the stationary reference frame and the resonant term is added to the main
controller to suppress the harmonic component in the specific order. However, since this method
requires separate resonant terms to compensate each harmonic component, this approach is usually
considered to alleviate only a few harmful low order harmonics. When the numbers of resonant terms
increase, the control structure becomes complicated or even impractical.
Other approaches to eliminate the harmonics from inverter current use nonlinear control
techniques such as the sliding mode control (SMC) [12], predictive control [13], or repetitive control [14].
These control strategies are often referred as non-selective method since the controllers work in wide
range of frequency, in contrast to the selective method that only regulates the harmonics in some
specified orders. By using such nonlinear controllers, the distorted level of inverter current can be
mitigated even under the distorted grid voltages. However, the design task of a robust controller
based on above techniques usually makes the system structure complicated because of the remaining
problems related to those techniques such as the chattering problem in SMC, parameter sensitive
in predictive control, slow dynamic response in repetitive control and so on. Furthermore, the
practical complexity of nonlinear controller may degrade the performance of the controlled system.
As another approach, a neural-network-based waveform processing and filtering scheme has been
reported to reshape voltage or current waveforms [15]. However, this algorithm generally requires
lots of computations to be processed in real time, which increases the computational burden of main
controller. Moreover, offline training is often required in this method.
Recently, a sliding mode harmonic compensation strategy based on the system model
decomposition has been reported to reduce the harmful effects caused by the nonlinear controller [16].
In this work, the system model is first divided into two using the fundamental and harmonic
components. Using two decomposed models, the controllers are separately designed, that is, the
controller for the fundamental term by the conventional decoupling controller and the harmonic
suppression controller by SMC. To decompose the grid voltages and inverter currents into the
fundamental and harmonic components, the fourth-order band pass filters (BPFs) have been employed
in the harmonic extractors. As reported in [16], even though the steady-state current responses can
be quite improved with reduced chattering by adopting the decomposed model, the inverter system
exhibits a slow transient response due to the sluggish dynamic characteristics of the BPF. Moreover,
the slow response of the BPF may even cause the instability problem during transient duration.
Energies 2016, 9, 190 3 of 23

In addition to current controller, the phase lock loop (PLL) also influences on the inverter currents
under the distorted grid voltages. Indeed, the conventional synchronous reference frame phase lock
loop (SRF-PLL) algorithm integrated with a PI controller to determine the angular displacement of grid
voltage is unable to cope with high frequency disturbances due to the limitation of the PI controller.
Consequently, the high frequency uncertainty caused by the distorted grid voltages has a profound
effect on the determination of the angular displacement of grid voltages. The effects of distorted grid
on PLL can be alleviated by reducing the PLL bandwidth. However, this results in slow dynamic
response in PLL, which means that the PLL cannot track the angular displacement of grid voltages
rapidly. Recent studies have proved the effectiveness of the moving average filter basis PLL (MAF-PLL)
under the distorted grid voltages [17,18]. The main concept of the MAF-PLL is to use the MAF as an
ideal low pass filter (LPF) to remove the sinusoidal components in the synchronous reference frame
before the measured grid voltages are processed in PLL algorithm.
The main objective of this paper is to present a robust current control scheme for a three-phase
grid-connected inverter under abnormal grid conditions like the distorted grid. The proposed control
strategy is based on the system model decomposition, in which the fundamental and harmonic current
controllers are designed separately. Whereas the synchronous PI decoupling controller is employed to
control the fundamental current component, a predictive basis compensator is introduced to suppress
the harmonic components in inverter currents. For this purpose, the harmonic contents are extracted
by means of the MAFs. Furthermore, an MAF-PLL is employed to improve the detecting performance
of the conventional SRF-PLL. Also, a simple modification method to improve the transient current
response is presented by changing the q-axis and d-axis harmonic currents only during the transient
period of the MAF. As a result, the proposed control scheme can effectively control a grid-connected
inverter during steady-state as well as transient periods even under the distorted grid voltages.
This paper is organized as follows: Section 2 presents the mathematical system model as well
as the decomposition method. Section 3 describes the proposed control scheme composed of the PI
decoupling controller, MAF-based harmonic extractor, MAF-PLL, and the predictive basis harmonic
compensator. Simulation results are presented in Section 4. Afterwards, the experimental results are
provided in Section 5. Finally, the conclusions are given in Section 6.

2. Modeling of a Grid-Connected Inverter

Figure 1 shows a whole configuration of three-phase grid-connected inverter with an L filter.
The analyses for current controller in inverter with an L filter are still valid for the case with LCL filter
as long as the control bandwidth is kept below resonant frequency of LCL filter [19]. Considering the
filter inductance of L and the filter resistance of R, the voltage equations of a grid-connected inverter
are expressed in the synchronous reference frame as follows:

diq
vq “ Riq ` L ` ωLid ` eq (1)
dt
did
vd “ Rid ` L ´ ωLiq ` ed (2)
dt
where iq and id are the q-axis and d-axis inverter currents, respectively; eq and ed are the q-axis and d-axis
grid voltages, respectively; vq and vd are the q-axis and d-axis inverter output voltages, respectively;
and ω is the grid angular frequency.
 𝑑𝑖𝑑
𝑣𝑑 = 𝑅𝑖𝑑 + 𝐿 − ω𝐿𝑖𝑞 + 𝑒𝑑 (2)
𝑑𝑡
where 𝑖𝑞 and 𝑖𝑑 are the q-axis and d-axis inverter currents, respectively; 𝑒𝑞 and 𝑒𝑑 are the q-axis
and d-axis grid voltages, respectively; 𝑣𝑞 and 𝑣𝑑 are the q-axis and d-axis inverter output voltages,
Energies 2016, 9, 190 4 of 23
respectively; and 𝜔 is the grid angular frequency.

Figure 1. Block diagram of three-phase grid-connected inverter.

Figure 1. Block diagram of three-phase grid-connected inverter.

When phase voltages and currents include harmonic components, the transformed variables into
the synchronous reference frame are not in pure direct current (DC)-quantity. Instead, they contain the
harmonic components as well as DC-quantity. To apply the decomposition method, the voltage and
current variables can be expressed with DC-quantity and sinusoidal harmonic terms as follows:

vqd “ Vqd ` vqdh (3)

iqd “ Iqd ` iqdh (4)

eqd “ Eqd ` eqdh (5)

where the capital letter V, I, and E denote the inverter voltages, inverter currents, and grid voltages
having DC-quantity, respectively, the subscript “qd” denotes the variables on the q-axis and d-axis in
the synchronous reference frame, the subscripts “h” denotes harmonic quantities.
The decomposed models for the voltage equations of a grid-connected inverter are derived by
substituting Equations (3), (4), and (5) into Equations (1) and (2) as follows:
´ ¯ dpIq ` iqh q ´ ¯
Vq ` vqh “ R Iq ` iqh ` L ` ωL pId ` idh q ` Eq ` eqh (6)
dt
dpId ` idh q ´ ¯
Vd ` vdh “ R pId ` idh q ` L ´ ωL Iq ` iqh ` pEd ` edh q (7)
dt
Voltage equations in Equations (6) and (7) can be rewritten with respect to the DC-quantities and
harmonic components as follows:

dIqs
Vq “ RIq ` L ` ωLId ` Eq (8)
dt
dId
Vd “ RId ` L ´ ωLIq ` Ed (9)
dt
diqh
vqh “ Riqh ` L ` ωLidh ` eqh (10)
dt
didh
vdh “ Ridh ` L ´ ωLiqh ` edh (11)
dt
For convenience, Equations (8) and (9) are referred to the fundamental voltage equations because
they are obtained through the Park’s transformation of phase variables in the fundamental components.
Similarly, Equations (10) and (11) are referred to the harmonic voltage equations.
Energies 2016, 9, 190 5 of 23

3. Proposed Control Scheme

The issue on power quality of DG systems is one of most common interconnection requirements
for grid-connected inverter basis generation systems, which is being presented in all standards.
According to the published standards [6,7], the grid injected current should not have the total harmonic
distortion (THD) greater than 5% in most of operating conditions. Although recently developed current
controllers for grid-connected inverters including linear and non-linear schemes can fulfill the THD
Energies 2016, 9, 190 5 of 21
requirement in normal grid condition, the THD of inverter output current may exceed the limit during
distorted grid voltages. The main purpose of the proposed control scheme is not only to fulfill the grid
voltage accurately. Similarly, the transformed currents into the synchronous frame are processed
interface requirements but also to minimize the harmonic contents in grid injected currents.
with the MAF to obtain DC-quantity. These voltages and currents in DC-quantity are subtracted from
Figure 2 shows the control block diagram of the proposed control scheme in which the three-phase
𝑒𝑞𝑑 and 𝑖𝑞𝑑 to extract the harmonic components of 𝑒𝑞𝑑ℎ and 𝑖𝑞𝑑ℎ . The current controller is formed
inverter is connected to the grid through L filter. The overall system mainly consists of a three-phase
by combining a synchronous PI decoupling controller and a predictive controller, where the PI
inverter, a current controller, a predictive basis harmonic compensator, and MAFs, which are used
decoupling controller is employed to regulate the inverter currents to follow their references and
for system model decomposition as well as performance improvement in PLL. The grid voltages
referred as a main controller.
and inverter currents are measured and then transferred into the variables on the synchronous
From the inverter voltage equations in Equations (1) and (2), the reference voltages are obtained
reference frame using the Park’s transformation. The transformed grid voltages into the synchronous
from the PI decoupling controller as follows:
frame are first processed with the MAF to obtain pure DC-quantity without the harmonic distortion,
and then, used in PLL algorithm𝑉to ∗ determine
𝑘𝑖 the∗angular displacement in grid voltage accurately.
𝑞 = (𝑘𝑝 + ) ∙ (𝑖𝑞 − 𝑖𝑞 ) + 𝜔𝐿𝐼𝑑 + 𝐸𝑞 (12)
𝑠
Similarly, the transformed currents into the synchronous frame are processed with the MAF to obtain
DC-quantity. These voltages and currents 𝑘𝑖
in DC-quantity are subtracted from eqd and iqd to extract the
𝑉𝑑∗ = (𝑘𝑝 + ) ∙ (𝑖𝑑∗ − 𝑖𝑑 ) − 𝜔𝐿𝐼𝑞 + 𝐸𝑑 (13)
𝑠
harmonic components of eqdh and iqdh . The current controller is formed by combining a synchronous
PI decoupling
where controller
the symbol “*” and a predictive
denotes controller,
the reference the PI𝑘decoupling
where and,
quantity 𝑝 and 𝑘𝑖 are the proportional and
controller is employed to
regulate the inverter currents to follow their
integral gains of the PI controller, respectively. references and referred as a main controller.

Figure2.2. Block
Figure Blockdiagram
diagramof
ofthe
theproposed
proposedcontrol
controlscheme.
scheme.

In thethe
From discrete-time domain,
inverter voltage the reference
equations voltages
in Equations (1) in
andEquations (12) and voltages
(2), the reference (13) can are
be expressed
obtained
at each sampling instant k as follows:
from the PI decoupling controller as follows:
ˆ ˙ ´𝑘
𝑉𝑞∗ (𝑘) k
= 𝑘V “ (𝑘)
𝑝 q∙˚𝑖𝑞,𝑒𝑟𝑟 k p `+ 𝑘𝑖 ∙ ∑i˚q 𝑖´ iq (𝑖)
` ∙ωLI
𝑇𝑠 +d `
ω𝐿𝐼
Eq𝑑 (𝑘) + 𝐸𝑞 (𝑘)
¯
i (14)
𝑞,𝑒𝑟𝑟 (12)
s 𝑖=0
ˆ ˙ 𝑘
k
𝑉𝑑∗ (𝑘) k p `+ 𝑘i𝑖
`˚ ˘
=
˚
“ (𝑘)
𝑘𝑝V∙d 𝑖𝑑,𝑒𝑟𝑟 ∙∑id 𝑖´ id (𝑖)
𝑑,𝑒𝑟𝑟 ∙ 𝑇𝑠 −
´ ωLI q`ω𝐿𝐼
Ed𝑞 (𝑘) + 𝐸𝑑 (𝑘) (13)
(15)
s
𝑖=0
where the symbol “*” denotes the reference quantity and, k p and k i are the proportional and integral
where 𝑖𝑞,𝑒𝑟𝑟 = 𝑖𝑞∗ − 𝑖𝑞 , 𝑖𝑑,𝑒𝑟𝑟 = 𝑖𝑑∗ − 𝑖𝑑 , and Ts is the sampling period.
gains of the PI controller, respectively.
3.1. MAF and MAF-PLL
An MAF is employed as ideal LPF in the proposed control scheme, where the window length is
chosen as one period of the fundamental grid voltage to retain the generality. However, the window
length can be set to half of fundamental period in the case of the filtered variables are in the
synchronous reference frame since the majority of the uncertainties in three-phase voltages are odd
harmonics that become even harmonics in the synchronous reference frame.
Energies 2016, 9, 190 6 of 23

In the discrete-time domain, the reference voltages in Equations (12) and (13) can be expressed at
each sampling instant k as follows:

k
ÿ
Vq˚ pkq “ k p iq,err pkq ` k i iq,err piq Ts ` ωLId pkq ` Eq pkq (14)
i “0

k
ÿ
Vd˚ pkq “ k p id,err pkq ` k i id,err piq Ts ´ ωLIq pkq ` Ed pkq (15)
i “0

where iq,err “ i˚q ´ iq , id,err “ i˚d ´ id , and Ts is the sampling period.

3.1. MAF and MAF-PLL

An MAF is employed as ideal LPF in the proposed control scheme, where the window length is
chosen as one period of the fundamental grid voltage to retain the generality. However, the window
length can be set to half of fundamental period in the case of the filtered variables are in the synchronous
reference frame since the majority of the uncertainties in three-phase voltages are odd harmonics that
become even harmonics in the synchronous reference frame.
The input–output relationship of the MAF can be described in continuous-time domain by
żt
1
xout ptq “ xin pτq dτ (16)
Tw t´Tw

where Tw is the window length, xout ptq is the output signal, and xin pτq is the input signal. The
continuous-time transfer function of the MAF can be obtained from Equation (16) as

xout psq 1 ´ e´Tw s

G MAF psq “ “ (17)
xin psq Tw s

The transfer function in Equation (17) shows that the MAF only requires a time equal to its
window length to reach the steady state. For a digital implementation, the transfer function of the
MAF can be discretized from Equation (17) as

1 1 ´ z´ N
G MAF pzq “ (18)
N 1 ´ z ´1

where Tw = NTs with N equal to the number of samples in one window length (N must be an integer).
From Equation (18), it can be observed that the MAF provides an effective solution in view of the
computational burden as compared with the BPF. Moreover, since the MAF only requires one period
of fundamental grid voltage for the filter output to converge, the dynamic response of the MAF-based
harmonic extractor is much faster than that of the BPF-based harmonic extractor, which requires almost
six periods of fundamental grid voltage to reach steady-state condition [16].
In practical implementation, the mismatch between the designed window length Tw and the value
of NTs makes errors in amplitude and phase angle. This problem can be solved by using several proved
techniques such as rounding of Tw /Ts to the nearest integer, weighted mean value approach [20],
and linear interpolation [21]. In order to secure the calculating capability of main controller while
retaining the errors within an acceptable value [17], the rounding scheme that adjusts Tw /Ts to the
nearest integer is used in this paper. Thus, the number of samples in one window length N can be
obtained as follows: ˆ ˙
Tw
N “ round (19)
Ts
where “round” denotes the round function which returns the nearest integral value.
Energies 2016, 9, 190 7 of 23

In the grid-connected inverter system, the inverter needs to monitor the angular displacement of
the grid voltage in order to synchronize the inverter currents with the grid. Generally, the SRF-PLL,
which is based on the Park’s transformation and a low bandwidth PI controller, is commonly employed
in three-phase grid-tied inverter systems. Under the distorted grid voltages, however, the angular
displacement of grid voltage cannot be tracked exactly due to the inability of the PI controller to
reject sinusoidal disturbances. As a result, the detected angular frequency of grid contains high order
harmonics, which cause an increase of harmonic contents in inverter currents. As is illustrated in
Figure 2, the filtered d-axis voltage is fed directly to the conventional SRF-PLL instead of the measured
d-axis voltage.
The open loop transfer function of MAF-PLL can be obtained from Equation (17) and the
conventional SRF-PLL as
k p,pll s ` k i,pll
Gol psq “ G MAF psq ˚ (20)
s2
where k p,pll and k i,pll are the proportional and integral gains of the PI controller in the conventional
SRF-PLL, respectively. By using the first-order Padé approximation [17], Equation (17) may be
approximated as
1
G MAF psq « (21)
Tw
s`1
2
Replacing G MAF psq in Equation (20) by approximated term G MAF psq in Equation (21), the transfer
function in Equation (20) becomes
´ ¯
2 k p,pll s ` k i,pll
G MAF psq « (22)
pTw s ` 2q s2

From Equation (22), the approximated closed-loop transfer function of MAF-PLL can be
described as
k p,pll k i,pll
2 s`2
Tw Tw
Gcl psq « (23)
2 k p,pll k i,pll
3
s ` 2
s `2 s`2
Tw Tw Tw
By using the MAF, which acts as an ideal LPF, to obtain DC-quantity of the d-axis grid voltage, the
grid angular displacement can be detected accurately even when the grid voltages are highly distorted.
In addition to mathematical analyses, the simulation and experimental results of MAF-PLL will be
given in the following sections to demonstrate the performance of the designed filter as well as its
effect on inverter currents.

3.2. Predictive Harmonic Compensator

The harmonic compensator, which is based on the predictive control, is also illustrated in Figure 2.
The high frequency sinusoidal currents and voltages are extracted by means of the MAFs, and then,
these values are used to form a predictive basis harmonic compensator. In the discrete-time domain,
the harmonic voltage equations in Equations (10) and (11) can be described as

L ´ ¯
vqh pkq “ Riqh pkq ` iqh pk ` 1q ´ iqh pkq ` ωLidh pkq ` eqh pkq (24)
Ts

L
vdh pkq “ Ridh pkq ` pi pk ` 1q ´ idh pkqq ´ ωLiqh pkq ` edh pkq (25)
Ts dh
Energies 2016, 9, 190 8 of 23

In Equations (24) and (25), by replacing the currents at sampling instant pk ` 1) by the reference
currents, the predictive control law to suppress the harmonic currents can be expressed as

L ´˚ ¯
v˚qh pkq “ Riqh pkq ` iqh pkq ´ iqh pkq ` ωLidh pkq ` eqh pkq (26)
Ts

L `˚ ˘
v˚dh pkq “ Ridh pkq ` idh pkq ´ idh pkq ´ ωLiqh pkq ` edh pkq (27)
Ts
where i˚qh pkq and i˚dh pkq denote the q-axis and d-axis harmonic reference currents, respectively. These
values should be set to zeros in order to eliminate the harmonic components in inverter output current.
From Equations (14), (15), (26), and (27) with i˚qh pkq “ i˚dh pkq “ 0, the q-axis and d-axis inverter
reference voltages of the proposed control scheme are represented as

v˚q pkq “ Vq˚ pkq ` v˚qh pkq (28)

v˚d pkq “ Vd˚ pkq ` v˚dh pkq (29)

As a result of using the decomposition model, the predictive basis harmonic compensator only has
to deal with the harmonics model. Therefore, the problems associated with the parameters mismatch
and time delay in the predictive control are highly reduced.

4. Simulation Results
To evaluate the performance of the proposed control scheme, the simulations have been done
for three-phase grid-connected inverter using the PSIM software. Figure 3 shows the whole system
configuration where the main controller is implemented with the PSIM dynamic link library (DLL)
block.
EnergiesThe
2016,system
9, 190 parameters are summarized in Table 1. 8 of 21

Figure3.3. The
Figure The whole
whole system
systemconfiguration
configurationfor
forsimulation.
simulation.

Table 1. System parameters of grid-connected inverter.

Parameters Symbol Value Units

Rated power PR 2 kW
Grid voltage eg 180 V
Grid frequency f 60 Hz
DC-link voltage VDC 420 V
Filter inductance L 7 mH
Filter resistance R 0.5 Ω
Switching frequency fs 10 kHz

In addition to ideal grid voltages, the distorted grid voltages are used in order to evaluate the
control performance in adverse operating conditions of the grid. For a distorted condition of grid
voltage, the 5th and 7th harmonics with 20% of the fundamental component and the 11th and 13th
Energies 2016, 9, 190 9 of 23
Figure 3. The whole system configuration for simulation.

Table1.1.System
Table Systemparameters
parametersofofgrid-connected
grid-connectedinverter.
inverter.

Parameters
Parameters Symbol Value
Symbol
Value Units Units
Rated power PR 2 kW
Rated power PR 2 kW
Grid voltageGrid voltage eg eg 180
180 V V
Grid frequencyGrid frequency f f 60
60 Hz Hz
DC-link voltage VDC VDC
DC-link voltage 420
420 V V
Filter inductance L
Filter inductance L 77 mH mH
Filter resistance R 0.5 Ω
Filter resistance R 0.5 Ω
Switching frequency fs 10 kHz
Switching frequency fs 10 kHz

In
Inaddition
additiontotoideal
idealgrid
gridvoltages,
voltages,the
thedistorted
distortedgrid
gridvoltages
voltagesare
areused
usedin inorder
ordertotoevaluate
evaluatethe the
control performance in adverse operating conditions of the grid. For a distorted condition
control performance in adverse operating conditions of the grid. For a distorted condition of grid of grid
voltage,
voltage,the
the5th
5thand
and7th7thharmonics
harmonicswithwith20%
20%ofofthe
thefundamental
fundamentalcomponent
componentand andthe
the11th
11thand and13th
13th
harmonics with 10% of the fundamental component are added to the ideal grid
harmonics with 10% of the fundamental component are added to the ideal grid voltages. Thevoltages. The resultant
three-phase grid voltages
resultant three-phase gridare shownare
voltages in Figure
shown4,inwhere
Figurethe THD isthe
4, where 31.7%.
THDIn is this figure,
31.7%. ea , figure,
In this eb , and 𝑒ec ,
𝑎
denote three-phase grid voltages.
𝑒 , and 𝑒 denote three-phase grid voltages.
𝑏 𝑐

Figure4.4.Distorted
Figure Distortedthree-phase
three-phasegrid
gridvoltages.
voltages.

Figure55shows
Figure showsthe
theperformance
performancecomparison
comparisonof ofdynamic
dynamicresponses
responsesbetween
betweenthe theBPF
BPFand
andMAF
MAF
under the distorted grid voltages. As a result of harmonic distortion in grid voltages as shown inin
under the distorted grid voltages. As a result of harmonic distortion in grid voltages as shown
Figure4,4,the
Figure theq-axis
q-axisand
andd-axis
d-axisgrid
grid
voltages e 𝑒and
voltages 𝑞 and
q
𝑒𝑑the
e at at synchronous
d
the synchronous reference
reference frameframe contain
contain the
harmonic components as well as DC-quantity. In Figure 5, eq,BPF and ed,BPF denote the filtered outputs
processed with the BPF, respectively. On the other hand, eq,MAF and ed,MAF denote the filtered outputs
processed with the MAF, respectively. It is easy to notice that the transient responses of the MAF are
nearly five times faster than those of the BPF in extraction of DC-quantity, since the MAF only requires
one window length to reach its steady state. Furthermore, by selecting the number of samples N to the
nearest integer, the steady-state errors of the MAF can be lower than those of the fourth-order BPF
(See the d-axis values).
 of the MAF are nearly five times faster than those of the BPF in extraction of DC-quantity, since the
MAF only requires one window length to reach its steady state. Furthermore, by selecting the number
of samples N to the nearest integer, the steady-state errors of the MAF can be lower than those of the
fourth-order BPF (See the d-axis values).
Similarly, Figure 6 shows the comparison of dynamic responses to extracts DC-quantity between
the BPF and MAF when the input signals are inverter currents. It is obvious that the transient
Energies 2016, 9, 190 10 of 23
response and steady-state error are much more improved by adopting the MAF, which results in the
improvement of harmonic extraction.

Figure 5. Performance comparison of dynamic responses between the band pass filter (BPF) and
Figure 5. Performance comparison of dynamic responses between the band pass filter (BPF) and
moving average filter (MAF) when direct current (DC)-quantity of grid voltages is extracted at the
moving average filter
synchronous (MAF)
frame underwhen directgrid
the distorted current (DC)-quantity
voltages: of(b)
(a) with BPF; and grid
withvoltages
MAF. is extracted at the
synchronous frame under the distorted grid voltages: (a) with BPF; and (b) with MAF.

Similarly, Figure 6 shows the comparison of dynamic responses to extracts DC-quantity between
the BPF and MAF when the input signals are inverter currents. It is obvious that the transient
response and steady-state error are much more improved by adopting the MAF, which results in the
improvement of harmonic extraction.
Energies 2016, 9, 190 10 of 21

Figure 6. Performance comparison of dynamic responses between the BPF and MAF when
Figure 6. Performance comparison of dynamic responses between the BPF and MAF when DC-quantity
DC-quantity of inverter currents is extracted at the synchronous frame under the distorted grid voltages:
of inverter(a)currents is and
using BPF; extracted
(b) usingatMAF.
the synchronous frame under the distorted grid voltages: (a) using
BPF; and (b) using MAF.
As stated earlier, the obtained angular displacement and angular frequency from the
conventional SRF-PLL are severely distorted under the distorted grid voltages because of the inability
of the PI controller to reject high frequency disturbances. Figure 7 shows the waveforms for the
angular frequency 𝜔PLL and angular displacement θ determined by the conventional SRF-PLL. It is
easy to notice that the angular frequency fluctuates around the fundamental value 𝜔𝑔𝑟𝑖𝑑 . Also, this
fluctuation makes the obtained angular displacement θ distorted as is shown in Figure 7b. On the
contrary, Figure 8 shows the simulation results for the MAF-PLL scheme under the distorted grid
voltages. It can be clearly observed that the dynamic response of the MAF-PLL is comparable to that
achieved by the SRF-PLL. The MAF-PLL, however, provides attractive steady-state response as a
Energies 2016, 9, 190 11 of 23
Figure 6. Performance comparison of dynamic responses between the BPF and MAF when
DC-quantity of inverter currents is extracted at the synchronous frame under the distorted grid voltages:
As stated earlier,
(a) using BPF;the
andobtained angular displacement and angular frequency from the conventional
(b) using MAF.
SRF-PLL are severely distorted under the distorted grid voltages because of the inability of the PI
As stated earlier, the obtained angular displacement and angular frequency from the
controller to reject high frequency disturbances. Figure 7 shows the waveforms for the angular
conventional SRF-PLL are severely distorted under the distorted grid voltages because of the inability
frequency ω
of thePLL
and angular displacement θ determined by the conventional SRF-PLL. It is easy to
PI controller to reject high frequency disturbances. Figure 7 shows the waveforms for the
notice that
angular frequencyfrequency
the angular fluctuates
𝜔PLL and angular around the
displacement fundamental
θ determined value
by the ω grid . Also,
conventional this It
SRF-PLL. fluctuation
is
makes the
easy obtained angular
to notice that displacement
the angular θ distorted
frequency fluctuates as the
around is shown
fundamental value 𝜔
in Figure 7b.
𝑔𝑟𝑖𝑑 .On the
Also, thiscontrary,
Figure 8fluctuation
shows the makes the obtained
simulation angular
results displacement
for the MAF-PLL θ distorted as is shown
scheme under in Figure 7b.
the distorted Onvoltages.
grid the It
contrary, Figure 8 shows the simulation results for the MAF-PLL scheme under
can be clearly observed that the dynamic response of the MAF-PLL is comparable to that achieved the distorted grid
voltages. It can be clearly observed that the dynamic response of the MAF-PLL is comparable to that
by the SRF-PLL. The MAF-PLL, however, provides attractive steady-state response as a result of an
achieved by the SRF-PLL. The MAF-PLL, however, provides attractive steady-state response as a
excellent filtering
result capability
of an excellent of the
filtering MAF. of the MAF.
capability

Figure Figure 7. Dynamic response of the conventional synchronous reference frame phase lock loop
7. Dynamic response of the conventional synchronous reference frame phase lock loop
(SRF-PLL) under the distorted grid voltages: (a) angular frequency; and (b) angular displacement.
(SRF-PLL) under the distorted grid voltages: (a) angular frequency; and (b) angular displacement.
Energies 2016, 9, 190 11 of 21

Figure 8. Dynamic response of the moving average filter phase lock loop (MAF-PLL) under the
Figure 8. Dynamic response of the moving average filter phase lock loop (MAF-PLL) under the
distorted grid voltages: (a) angular frequency; and (b) angular displacement.
distorted grid voltages: (a) angular frequency; and (b) angular displacement.
To highlight the effectiveness of the proposed control scheme, the comparative simulation
results withthe
To highlight the effectiveness
conventional control
of the scheme for acontrol
proposed grid-connected
scheme,inverter are also presented.
the comparative simulation The results
conventional control scheme comprises a synchronous PI decoupling current controller and a SRF-
with the conventional control scheme for a grid-connected inverter are also presented. The conventional
PLL. In order to point out the weakness of the conventional control scheme, the simulation results
control both
scheme comprises
under a synchronous
the ideal and distorted grid PI decoupling
voltages current
are shown controller
in Figures and
9 and 10, a SRF-PLL.
in which In order to
𝑖𝑎 , 𝑖𝑏 , and
𝑖𝑐 denote three-phase inverter currents. As can be observed from the output current waveforms in
Figure 9, the conventional control scheme shows a good steady-state performance under the ideal
grid voltages. In the abnormal grid conditions, however, the output current waveforms, illustrated
in Figure 10, are highly distorted due to the poor sinusoidal disturbance rejection capability of the PI
controller both in current controller and in PLL.
 Figure 8. Dynamic response of the moving average filter phase lock loop (MAF-PLL) under the
distorted grid voltages: (a) angular frequency; and (b) angular displacement.
Energies 2016, 9, 190 12 of 23
To highlight the effectiveness of the proposed control scheme, the comparative simulation
results with the conventional control scheme for a grid-connected inverter are also presented. The
point out the weakness
conventional of scheme
control the conventional
comprises a control scheme,
synchronous the simulation
PI decoupling currentresults
controllerboth
andunder
a SRF-the ideal
and distorted grid voltages are shown in Figures 9 and 10 in which
PLL. In order to point out the weakness of the conventional control scheme, i , i , and i
a b the simulation
c denote three-phase
results
both under the ideal and distorted grid voltages are shown in Figures 9 and
inverter currents. As can be observed from the output current waveforms in Figure 9, the 10, in which 𝑖 𝑎 , 𝑖 , and
𝑏conventional
𝑖 denote
control scheme
𝑐 shows a good steady-state performance under the ideal grid voltages. In the in
three-phase inverter currents. As can be observed from the output current waveforms abnormal
Figure 9, the conventional control scheme shows a good steady-state performance under the ideal
grid conditions, however, the output current waveforms, illustrated in Figure 10, are highly distorted
grid voltages. In the abnormal grid conditions, however, the output current waveforms, illustrated
due to the poor sinusoidal disturbance rejection capability of the PI controller both in current controller
in Figure 10, are highly distorted due to the poor sinusoidal disturbance rejection capability of the PI
and in PLL.
controller both in current controller and in PLL.

Figure Figure 9. Inverter

9. Inverter currents
currents of the
of the conventional proportional-integral
conventional proportional-integral (PI) (PI)
decoupling controller
decoupling under under
controller
the ideal grid voltages: (a) phase-currents; and (b) q-axis and d-axis currents.
the2016,
Energies ideal 190 voltages: (a) phase-currents; and (b) q-axis and d-axis currents.
9, grid 12 of 21

Figure
Figure 10. Inverter currents
10. Inverter currents of
of the
the conventional PI decoupling
conventional PI controller under
decoupling controller under the
the distorted
distorted grid
grid
voltages: (a) phase-currents; and (b) q-axis and d-axis currents.
voltages: (a) phase-currents; and (b) q-axis and d-axis currents.

In addition to the results obtained using the conventional PI control scheme, the simulations
have been carried out for another harmonic compensation approaches for comparison. Figure 11
shows the steady-state responses of the inverter currents using PR controller with different distorted
grid conditions. To eliminate harmonic contents in inverter currents, only the resonant terms to
compensate the 5th and 7th harmonics are added together with the fundamental controller. As shown
in Figure 11a, the compensation works well when the grid disturbance only contains 5th and 7th
Energies 2016, 9, 190 13 of 23

Figure 10. Inverter currents of the conventional PI decoupling controller under the distorted grid
In addition to (a)
voltages: the results obtained
phase-currents; using
and (b) q-axis andthe conventional
d-axis currents. PI control scheme, the simulations
have been carried out for another harmonic compensation approaches for comparison. Figure 11
In addition to the results obtained using the conventional PI control scheme, the simulations
shows the steady-state responses of the inverter currents using PR controller with different distorted
have been carried out for another harmonic compensation approaches for comparison. Figure 11
grid conditions. To eliminate
shows the steady-state harmonic
responses contents
of the inverter in inverter
currents using PR currents,
controlleronly the resonant
with different terms to
distorted
compensate
grid conditions. To eliminate harmonic contents in inverter currents, only the resonant terms to shown
the 5th and 7th harmonics are added together with the fundamental controller. As
compensate
in Figure 11a, the the 5th and 7th harmonics
compensation works arewelladded
when together with the
the grid fundamental
disturbance controller.
only As shown
contains 5th and 7th
in Figure 11a, the compensation works well when the grid disturbance
harmonics. However, when the grid has the harmonic disturbance in other frequencies, only contains 5th andthe
7thcontrol
harmonics. However, when the grid has the harmonic disturbance in other frequencies, the control
performance of this scheme is degraded dramatically.
performance of this scheme is degraded dramatically.

Figure
Figure 11. 11. Inverter
Inverter currents
currents usingusing
PRPR controllerwith
controller with different
differentdistorted
distortedgridgrid
conditions: (a) The(a)
conditions: case
The case
when the grid disturbance only contains 5th and 7th harmonics. (b) The case when the grid
when the grid disturbance only contains 5th and 7th harmonics. (b) The case when the grid disturbance
disturbance contains 5th, 7th, 11th and 13th harmonics.
contains 5th, 7th, 11th and 13th harmonics.

Another widely used compensation approach that employs nonlinear controller is the sliding
mode control. As an illustration, Figure 12 shows the control performance under distorted grid
conditions when the sliding mode control is employed for harmonic compensation. As stated earlier,
this control scheme can guarantee a good steady-state performance of target system, as shown in
Figure 12a. However, the chattering in control input signals is unavoidable as is observed in the
waveforms of inverter reference voltages v˚q and v˚d in Figure 12b. Moreover, it is not easy to meet
the trade-off between the transient responses and steady-state performance in designing the sliding
mode controller.
conditions when the sliding mode control is employed for harmonic compensation. As stated earlier,
this control scheme can guarantee a good steady-state performance of target system, as shown in
Figure Energies 2016, 9, 190
12a. However, the chattering in control input signals is unavoidable as is observed 13 of 21
in the
∗ ∗
waveforms Another
of inverter reference voltages 𝑣𝑞
widely used compensation approachand 𝑣 𝑑 in
that employs nonlinear controller is the sliding meet
Figure 12b. Moreover, it is not easy to
the trade-off
Energies 2016, 9, between
190
mode control. Asthe
an transient
illustration,responses
Figure 12 and steady-state
shows performance
the control performance in designing
under the 14
distorted grid sliding
of 23
mode controller.
conditions when the sliding mode control is employed for harmonic compensation. As stated earlier,
this control scheme can guarantee a good steady-state performance of target system, as shown in
Figure 12a. However, the chattering in control input signals is unavoidable as is observed in the
waveforms of inverter reference voltages 𝑣𝑞∗ and 𝑣𝑑∗ in Figure 12b. Moreover, it is not easy to meet
the trade-off between the transient responses and steady-state performance in designing the sliding
mode controller.

Figure 12.
Figure 12. Control
Control performance
performance with with harmonic
harmonic compensation
compensation using
using the
the sliding
sliding mode
mode control
control under
under
distorted grid
distorted gridconditions:
conditions: (a)
(a) inverter
inverter phase
phase currents;
currents;and
and(b)
(b)inverter
inverterreference
referencevoltages.
voltages.
Figure 12. Control performance with harmonic compensation using the sliding mode control under
distorted grid conditions: (a) inverter phase currents; and (b) inverter reference voltages.
Figure13
Figure 13 shows
shows thethe simulation
simulation results
results of
of the
the proposed
proposed control
control scheme
scheme under
under the
the same
same distorted
distorted
grid voltages
grid voltagesFigureas Figure
13 shows
as Figure 10.
10. theIn contrast to
simulationtoresults
In contrast those in
thoseofinthe Figure
proposed
Figure 10, the
control
10, the phase
scheme
phase current waveforms
underwaveforms
current in
the same distorted Figure
in Figure 13a
13a still remain
grid sinusoidal.
voltages as Figure Also,
10. In as can
contrast be
to observed
those in in
Figure the
10, q-axis
the phase and d-axis
current
still remain sinusoidal. Also, as can be observed in the q-axis and d-axis current waveforms in current
waveforms inwaveforms
Figure in
Figure 13a
13b,still
the remain sinusoidal.
proposed controlAlso,scheme
as can begives
observed in the
nearly q-axis and
constant d-axis current
currents, which waveforms
imply in phase
that
Figure 13b, the proposed control scheme gives nearly constant currents, which imply that phase
currentsFigure
currents have13b,
have
the proposed
relatively
relatively low
control scheme
lowharmonic
harmonic gives nearly constant currents, which imply that phase
contents.
contents.
currents have relatively low harmonic contents.

Figure
Figure 13. 13. Inverter
Inverter currents
currents of of
thetheproposed
proposed control
control scheme
schemeunder the the
under distorted grid voltages:
distorted grid voltages:
(a) phase-currents; and (b) q-axis and d-axis currents.
(a) phase-currents; and (b) q-axis and d-axis currents.
Figure 13. Inverter currents of the proposed control scheme under the distorted grid voltages:
(a) phase-currents;
Despite and
the fact that (b)harmful
the q-axis and d-axis currents.
disturbances in three-phase grid are mainly sinusoidal harmonics,
any type of disturbance may occur in gird in extreme cases. Figure 14 shows the comparative current
responses in the presence of random disturbances in three-phase grid voltages. Random values are
generated as shown in Figure 14c and these values are added with grid voltages to generate random
disturbances in three-phase. From the comparative simulation results between the proposed control
scheme in Figure 14a and the conventional PI control scheme in Figure 14b, it is verified that the
proposed control scheme is robust against even random disturbance in grid voltage.
harmonics, any type of disturbance may occur in gird in extreme cases. Figure 14 shows the
comparative current responses in the presence of random disturbances in three-phase grid voltages.
Random values are generated as shown in Figure 14c and these values are added with grid voltages
to generate random disturbances in three-phase. From the comparative simulation results between
the proposed
Energies control scheme in Figure 14a and the conventional PI control scheme in Figure1514b,
2016, 9, 190 of 23it
is verified that the proposed control scheme is robust against even random disturbance in grid voltage.

Figure 14.
Figure 14. Steady-state
Steady-state current
current responses
responses with
with random
random limited-band
limited-band disturbances:
disturbances: (a)
(a) using
using the
the
proposed control
proposed control scheme;
scheme; (b)
(b) using
using the
the conventional
conventional PI
PI controller;
controller; and
and (c)
(c) waveform
waveform ofof random
random
disturbances in
disturbances in three-phase
three-phasegrid
gridvoltages.
voltages.

Although low harmonic contents in inverter currents at steady state can be achieved even under
Although low harmonic contents in inverter currents at steady state can be achieved even under
the distorted grid voltages by using the proposed control scheme, the transient response of the
the distorted grid voltages by using the proposed control scheme, the transient response of the
proposed control scheme becomes worse than that of the conventional control scheme. The
proposed control scheme becomes worse than that of the conventional control scheme. The degradation
degradation in transient response is mainly caused by the phase delay of filters used for the harmonic
in transient response is mainly caused by the phase delay of filters used for the harmonic extractor.
extractor. Figure 15 shows the transient responses of the conventional control scheme, in which
Figure 15 shows the transient responses of the conventional control scheme, in which Figure 15a
Figure 15a shows the waveforms of phase currents and Figure 15b shows the waveforms of the q-axis
shows the waveforms of phase currents and Figure 15b shows the waveforms of the q-axis and d-axis
and d-axis currents with the q-aixs current reference 𝑖𝑞∗ . As can be observed form Figure 15b, the
currents with the q-aixs current reference i˚q . As can be observed form Figure 15b, the conventional
conventional scheme responds rapidly to the change of reference currents. However, as is seen in
scheme responds rapidly to the change of reference currents. However, as is seen in Figure 16, the
Figure 16, the transient response of the proposed control scheme is relatively slow due to the delay
transient response of the proposed control scheme is relatively slow due to the delay introduced by
introduced by the harmonic extractors. The transient response in Figure 16b is almost ten times
the harmonic extractors. The transient response in Figure 16b is almost ten times slower than that of
slower than that of the conventional control scheme. This result is unacceptable to be used for a
the conventional control scheme. This result is unacceptable to be used for a current control loop in
current control loop in DG systems because they are supposed to provide very fast response.
DG systems because they are supposed to provide very fast response.
Energies 2016, 9, 190 16 of 23
Energies 2016, 9, 190 15 of 21
Energies 2016, 9, 190 15 of 21

Figure 15. Transient

Transientresponses
responsesof of
the the conventional PI decoupling
currentcurrent controller under the
Figure
Figure 15.
15. Transient responses of conventional PI decoupling
the conventional PI decoupling controller
current under the
controller distorted
under the
distorted
grid grid
voltages: voltages: (a) phase-currents; and
and (b) q-axis(b) q-axis and
d-axisandd-axis currents.
distorted grid (a) phase-currents;
voltages: (a) phase-currents; and and
(b) q-axis currents.
d-axis currents.

Figure 16. Transient responses of the proposed current control scheme under the distorted grid
Figure16.
Figure 16.Transient
Transient responses
responses of proposed
of the the proposed current
current controlcontrol
schemescheme under
under the the distorted
distorted grid
grid voltages:
voltages: (a) phase-currents; and (b) q-axis and d-axis currents.
voltages: (a) phase-currents; and (b) q-axis and d-axis
(a) phase-currents; and (b) q-axis and d-axis currents. currents.

Even though the major advantage of using the MAF in the harmonic extractor over BPF or LPF
Even though
though themajor
majoradvantage
advantageofofusingusing theMAF MAFininthe theharmonic
harmonic extractor over BPFor or LPF
is thatEven
the transientthe
period introduced by filter isthe
well reduced extractor
and predictable, over
this delay BPF
by the LPF
MAFis
is that the transient
that thebetransient period introduced
periodfurther
introduced by filter
by filter is well
is well the reduced
reduced and predictable,
and predictable, this delay by the MAF
should compensated in order to improve transient performancethis delay by
of current the MAF
controller.
should be
should be compensated
compensated further
further in
in order
order toto improve
improve the the transient
transient performance
performance of of current
current controller.
controller.
A compensation method for phase delay has been proposed in [18]. Although this method reduces
A compensation
A compensation methodmethod forfor phase
phase delay
delay has
has been
been proposed
proposed in in [18].
[18]. Although
Although this method
method reduces
reduces
the transient response of the MAF, the steady-state error is noticeably increased.thisBy considering that
the transient
the reference response
transient currents
responseareof the
of the MAF,
MAF, the the steady-state
steady-state error
error is noticeably
is noticeably increased. By considering that
the generally obtained from the voltage control increased.
loop or droop By considering
control loopthat
in
the reference
the reference currents
currents are
are generally
generally obtained
obtained fromfrom the
the voltage
voltage control
control loop
loop oror droop
droop control
control loop in
in
grid-connected inverter systems, it is always possible to detect the change in reference currents.loop
Based
grid-connected
grid-connected inverter
inverter systems,
systems,the it is always
it istransient possible
always possible to detect
to of
detect the change
the change in reference currents. Based
on this information, to improve response current control in reference
loop, currents.
the q-axis Based
and d-axis
on this information, to improve the transient response of current control loop, the q-axis and d-axis
harmonic currents 𝑖𝑞ℎ and 𝑖𝑑ℎ are temporally replaced during the transient period of the MAF with
harmonic currents 𝑖𝑞ℎ and 𝑖𝑑ℎ are temporally replaced during the transient period of the MAF with
 Energies
Energies 2016, 2016, 9, 190
9, 190 17 of 23 16 of 21

on this information, to improve the transient

𝑖𝑞ℎ,𝑡response 𝐼𝑞∗ current control loop, the q-axis and d-axis
= 𝑖𝑞 − of (30)
harmonic currents iqh and idh are temporally replaced during the transient period of the MAF with
𝑖𝑑ℎ,𝑡 = 𝑖𝑑 − ˚𝐼𝑑∗ (31)
iqh,t “ iq ´ Iq (30)
where 𝑖𝑞ℎ,𝑡 and 𝑖𝑑ℎ,𝑡 are the approximated q-axis and d-axis harmonic currents based on the
i “ id ´ Id˚ (31)
assumption that the fundamental currents candh,treach theirs reference values instantly.
where
Figure 17iqh,tshows
and idh,tthe
are the approximated
simulation q-axis and
results of d-axis harmonic currents
the proposed control based on the assumption
scheme with temporary
that the fundamental currents can reach theirs reference values instantly.
replacement of the harmonic currents during the transient period of the MAF, which is one window
Figure 17 shows the simulation results of the proposed control scheme with temporary
length. In order to show the transient response of the proposed control scheme, the q-axis reference
replacement of the harmonic currents during the transient period of the MAF, which is one window
current islength.
increased
In orderfrom 5 A the
to show to 10 A at tresponse
transient = 0.2 s, of
and
the then,
proposeddecreased to 7 Athe
control scheme, at tq-axis
= 0.3reference
s. As is clearly
observedcurrent
in Figure 17b, thefrom
is increased q-axis
5 Acurrent can
to 10 A at t =reach
0.2 s, its
andreference value to
then, decreased immediately,
7 A at t = 0.3 while
s. As isretaining
clearly observed in Figure 17b, the q-axis current can reach its reference
good harmonic suppression characteristics in phase currents. Although the replacement of the value immediately, while
harmonicretaining
currents good harmonic
during suppression
transient periodcharacteristics
of the MAF in phase currents. Although
may increase harmonic thecontents
replacementin inverter
of the harmonic currents during transient period of the MAF may increase harmonic contents in
current, the transient performance can be considerably improved. Moreover, the increase of the
inverter current, the transient performance can be considerably improved. Moreover, the increase
harmonicofcomponents in inverter in
the harmonic components currents
inverter can be negligible
currents because
can be negligible it lasts
because during
it lasts only
during only one
onewindow
length ofwindow
the MAF. length of the MAF.

Figure
Figure 17. 17. Transient
Transient responsesofof the
responses the proposed
proposedcurrent controlcontrol
current scheme under the distorted
scheme under grid
the voltages
distorted grid
with temporary replacement of the harmonic currents during transient period: (a) phase-currents; and
voltages with temporary replacement of the harmonic currents during transient period: (a) phase-
(b) q-axis and d-axis currents.
currents; and (b) q-axis and d-axis currents.
Figure 18 shows the fast Fourier transform (FFT) result of a-phase current using the proposed
Figure 18 shows
control the fast
scheme under Fourier grid
the distorted transform
voltages,(FFT) result
in which of a-phase
the harmonic current
limits using
according to thethe proposed
IEEE
control scheme under the distorted
1547 interconnection grid
of distributed voltages,
generation in which
regulations arethe
alsoharmonic
indicated forlimits according
comparison. As cantobethe IEEE
1547 interconnection of7th
seen, the 5th and distributed
harmonics aregeneration regulations
significantly reduced andare also
all the indicated
harmonic for comparison.
components are within As can
the limits.
be seen, the 5th and 7th harmonics are significantly reduced and all the harmonic components are
within the limits.
 Figure 18 shows the fast Fourier transform (FFT) result of a-phase current using the proposed
control scheme under the distorted grid voltages, in which the harmonic limits according to the IEEE
1547 interconnection of distributed generation regulations are also indicated for comparison. As can
Energies
be seen, the
2016, 9, 190
5th and 7th harmonics are significantly reduced and all the harmonic components
18 ofare
23
within the limits.

Figure18.
Figure 18.Fast
FastFourier
Fouriertransform
transform(FFT)
(FFT)result
result
of of a-phase
a-phase current
current using
using thethe proposed
proposed scheme
scheme under
under the
the distorted grid voltages with harmonic limits according to the IEEE
distorted grid voltages with harmonic limits according to the IEEE std. 1547.std. 1547.

Energies 2016, 9, 190 17 of 21

5. Experimental Results
5. Experimental Results
The proposed control scheme has been implemented in a 2 kVA three-phase grid-connected
inverter
Thesystem
proposed as shown
controlinscheme
Figure has
19. The
beenwhole control algorithms
implemented in a 2 kVA arethree-phase
implemented using 32-bit
grid-connected
floating-point DSP TMS320F28335 with clock frequency of 150 MHz [22].
inverter system as shown in Figure 19. The whole control algorithms are implemented using The sampling period
32-bitis
set to 100 µs inDSP
floating-point theTMS320F28335
experiments, which yields
with clock the switching
frequency of 150frequency
MHz [22]. of 10sampling
The kHz. Theperiod
intelligent
is set
power
to 100 µmodule (IPM) is employed
s in the experiments, for three-phase
which yields the switchinggrid-connected
frequency of inverter.
10 kHz. The The inverter power
intelligent phase
currents are detected
module (IPM) by the
is employed forHall-effect
three-phasedevices and are converted
grid-connected inverter. The through internal
inverter 12-bit A/D
phase currents are
converters.
detected by The inverter reference
the Hall-effect devices voltages are appliedthrough
and are converted using the symmetrical
internal space
12-bit A/D vector PWM
converters. The
method. The parameters
inverter reference voltagesofare
theapplied
experimental
using system are summarized
the symmetrical in Table
space vector PWM 1. For distorted
method. The
condition
parameters ofof
the
thegrid voltages, 10%
experimental of the
system arefifth and the seventh
summarized in Table harmonics, respectively,
1. For distorted andof1%
condition theofgrid
the
eleventh
voltages, and
10%the thirteenth
of the harmonics,
fifth and respectively,
the seventh harmonics, arerespectively,
injected to the andgrid
1%voltage, which results
of the eleventh and the in
the THD of 14.2%.
thirteenth harmonics, respectively, are injected to the grid voltage, which results in the THD of 14.2%.

Figure 19.
Figure 19. Experimental test setup.
Experimental test setup.

Figure 20 shows the experimental results for the SRF-PLL and MAF-PLL algorithms under the
Figure 20 shows the experimental results for the SRF-PLL and MAF-PLL algorithms under the
distorted grid voltages for comparison. As can been seen, the angular frequency waveform obtained
distorted grid voltages
by the SRF-PLL for comparison.
algorithm As candue
is highly distorted beentoseen, the angular
the existence frequency
of high waveform
frequency obtained
disturbances in
by the SRF-PLL algorithm is highly distorted due to the existence of high frequency disturbances
three-phase grid voltages. In contrast, the angular frequency obtained from the MAF-PLL algorithm in
three-phase
remains nearlygridconstant
voltages.inInthe
contrast, the angular
steady-state frequency
even when obtained
the grid from
voltages thehighly
are MAF-PLL algorithm
distorted. As a
remains
result, the angular displacement is not influenced by the distorted grid voltages, which provesAs
nearly constant in the steady-state even when the grid voltages are highly distorted. a
the
effectiveness and robustness of the MAF-PLL.
 Figure 19. Experimental test setup.

Figure 20 shows the experimental results for the SRF-PLL and MAF-PLL algorithms under the
distorted grid voltages for comparison. As can been seen, the angular frequency waveform obtained
by the SRF-PLL algorithm is highly distorted due to the existence of high frequency disturbances in
Energies 2016, 9, 190 19 of 23
three-phase grid voltages. In contrast, the angular frequency obtained from the MAF-PLL algorithm
remains nearly constant in the steady-state even when the grid voltages are highly distorted. As a
result, the angular displacement is not influenced by the distorted grid voltages, which proves the
robustness of
effectiveness and robustness of the
the MAF-PLL.
MAF-PLL.

Figure 20. Experimental results for the SRF-PLL and MAF-PLL under the distorted grid voltages.
Figure 20. Experimental results for the SRF-PLL and MAF-PLL under the distorted grid voltages.
Energies 2016, 9, 190 18 of 21

Figure 21 shows the experimental

experimental results for inverter
inverter output currents when
when the
the conventional
conventional
synchronous
synchronous PI decoupling controller
controller and SRF-PLL algorithm
algorithm are employed under the distorted grid
voltages. It is clearly observed in this figure that phase current waveforms are quite non-sinusoidal.
non-sinusoidal.
This
This indicates
indicatesthat
thatthe
theconventional
conventionalcontrol scheme
control failsfails
scheme to meet the high
to meet the power qualityquality
high power in a realinsystem
a real
where
systemthere
where exist sinusoidal
there disturbances.
exist sinusoidal disturbances.

Figure 21.
Figure 21. Experimental
Experimental results for inverter
results for inverter output
output currents
currents using
using the
the conventional
conventional PI
PI decoupling
decoupling
controller and
controller and SRF-PLL
SRF-PLL under
under the
the distorted
distorted grid
grid voltages.
voltages.

Figure 22 shows the experimental results for inverter output currents using the proposed control
Figure 22 shows the experimental results for inverter output currents using the proposed control
scheme under the same conditions as Figure 21. Due to an effective compensating capability of the
scheme under the same conditions as Figure 21. Due to an effective compensating capability of
proposed control scheme, the inverter phase currents shows considerably sinusoidal waveforms
the proposed control scheme, the inverter phase currents shows considerably sinusoidal waveforms
regardless of the high distortion level in the grid voltages.
regardless of the high distortion level in the grid voltages.
 controller and SRF-PLL under the distorted grid voltages.

Figure 22 shows the experimental results for inverter output currents using the proposed control
scheme under the same conditions as Figure 21. Due to an effective compensating capability of the
proposed
Energies 2016,control
9, 190 scheme, the inverter phase currents shows considerably sinusoidal waveforms
20 of 23
regardless of the high distortion level in the grid voltages.

Figure 22. Experimental results for inverter output currents using the proposed control scheme under
Figure 22. Experimental results for inverter output currents using the proposed control scheme under
the distorted grid voltages.
the distorted grid voltages.

As mentioned previously in Figure 16, the delay of the MAF might slow down the dynamic
As mentioned
response in inverterpreviously
currents. Inin Figure
order to 16,get
therid
delay of the
of this MAF might
problem, slow downtechnique
a modification the dynamicthat
response in inverter currents. In order to get rid of this problem, a modification
replaces the harmonic current with temporary values has been used. The transient current technique thatresponse
replaces
the harmonic
of the proposed current with
control temporary
scheme undervalues haschange
the step been used.
in theThe transient
current current
reference response of the
is demonstrated in
proposed control scheme under the step change in the current reference is demonstrated
Figure 23. As is shown, the q-axis inverter current instantly reaches its new reference value, in Figure
which23.
As is shown,
validates fastq-axis
the the inverter
dynamic current
response instantly
of the reaches
proposed its new
control reference value, which validates the
scheme.
fast dynamic
Energies response of the proposed control scheme.
2016, 9, 190 19 of 21

Figure 23. Step response of the inverter currents using the proposed control scheme under the
Figure 23. Step response of the inverter currents using the proposed control scheme under the distorted
distorted grid voltages.
grid voltages.

Figure 24 shows the experimental FFT result of a-phase inverter current using the proposed
Figure
control 24 shows
scheme under thethe distorted
experimental
grid FFT resultIt of
voltages. a-phaseobserved
is clearly inverter that
current using the proposed
the low-order harmonic
control scheme
components inunder the distorted
the fifth, seventh,grid voltages.and
eleventh, It isthirteenth
clearly observed
order that
are the low-order harmonic
considerably reduced.
components inall
Furthermore, thethe
fifth, seventh,components
harmonic eleventh, and thirteenth
well order
satisfy the are considerably
harmonic reduced.
limits of IEEE std. Furthermore,
1547 [6].
all the harmonic components well satisfy the harmonic limits of IEEE std. 1547 [6].
 distorted grid voltages.

Figure 24 shows the experimental FFT result of a-phase inverter current using the proposed
control scheme under the distorted grid voltages. It is clearly observed that the low-order harmonic
components
Energies in the fifth, seventh, eleventh, and thirteenth order are considerably reduced.
2016, 9, 190 21 of 23
Furthermore, all the harmonic components well satisfy the harmonic limits of IEEE std. 1547 [6].

Figure 24. FFT result of a-phase current using the proposed control scheme under the distorted grid voltages.
Figure 24. FFT result of a-phase current using the proposed control scheme under the distorted
grid voltages.
As a result, the proposed control scheme has a good disturbance rejection capability in steady
state as well as fast dynamic response.
As a result, the proposed control scheme has a good disturbance rejection capability in steady
state as well as fast dynamic response.
6. Conclusions
In this paper, an improved current control strategy for a grid-connected inverter even in the
6. Conclusions
presence of distorted grid conditions has been proposed for DG systems. The proposed control
In this paper, an improved current control strategy for a grid-connected inverter even in the
scheme is designed to overcome the limitation of the conventional PI type control schemes under
presence of distorted grid conditions has been proposed for DG systems. The proposed control scheme
adverse operation conditions of grid. Two significant enhancements in the proposed control scheme
is designed to overcome the limitation of the conventional PI type control schemes under adverse
are the implementation of a robust PLL algorithm and the additional harmonic suppression scheme.
operation conditions of grid. Two significant enhancements in the proposed control scheme are the
This is achieved by introducing the model decomposition method and MAF, by which the inverter
implementation of a robust PLL algorithm and the additional harmonic suppression scheme. This is
model can be decomposed into the fundamental and harmonic components. Based on the derived
achieved by introducing the model decomposition method and MAF, by which the inverter model can
models, the fundamental and harmonic currents are independently controlled by a decoupling PI
be decomposed into the fundamental and harmonic components. Based on the derived models, the
controller and a predictive basis controller, respectively. The principle and mathematical description
fundamental and harmonic currents are independently controlled by a decoupling PI controller and a
predictive basis controller, respectively. The principle and mathematical description of the proposed
control scheme have been analyzed to emphasize the simplicity of the control structure. To verify the
feasibility and effectiveness of the proposed control scheme, the whole control systems are constructed
using 32-bit floating-point DSP TMS320F28335. It has been demonstrated through the comparative
simulations and experimental results that the proposed control scheme is an effective way to improve
the performance of a grid-connected inverter under the distorted grid voltages.

Acknowledgments: This research was supported by Basic Science Research Program through the National
Research Foundation of Korea (NRF) funded by the Ministry of Education (NRF-2014R1A1A2056436). This
work was supported by the Human Resources Development of the Korea Institute of Energy Technology
Evaluation and Planning (KETEP) grant funded by the Korea government Ministry of Trade, Industry & Energy.
(NO. 20154030200720).
Author Contributions: Ngoc Bao Lai and Kyeong-Hwa Kim conceived the main concept of the control structure
and developed the entire system. Ngoc Bao Lai carried out the research and analyzed the numerical data with the
guidance from Kyeong-Hwa Kim. Ngoc Bao Lai and Kyeong-Hwa Kim collaborated to prepare the manuscript.
Conflicts of Interest: The authors declare no conflict of interest.
Energies 2016, 9, 190 22 of 23

References
1. Chicco, G.; Mancarella, P. Distributed multi-generation: A comprehensive view. Renew. Sustain. Energy Rev.
2009, 13, 535–551. [CrossRef]
2. Llaria, A.; Curea, O.; Jiménez, J.; Camblong, H. Survey on microgrids: Unplanned islanding and related
inverter control techniques. Renew. Energy 2011, 36, 2052–2061. [CrossRef]
3. Zamora, R.; Srivastava, A.K. Controls for microgrids with storage: Review, challenges, and research needs.
Renew. Sustain. Energy Rev. 2010, 14, 2009–2018. [CrossRef]
4. Lasseter, R.; Akhil, A.; Marnay, C.; Stephens, J.; Dagle, J.; Guttromson, R.; Meliopoulous, A.S.; Yinger, R.;
Eto, J. The CERTS microgrid concept, white paper on integration of distributed energy resources.
U.S. Department of Energy 2002. Available online: http://bnrg.eecs.berkeley.edu/~randy/Courses/
CS294.F09/MicroGrid.pdf (accessed on 15 January 2016).
5. Guerrero, J.M.; Vasquez, J.C.; Matas, J.; de Vicuña, L.G.; Castilla, M. Hierarchical control of droop—Controlled
AC and DC microgrids—A general approach toward standardization. IEEE Trans. Ind. Electr. 2011, 58,
158–172. [CrossRef]
6. IEEE Std. 1547-2003. Standard for interconnecting distributed resources with electric power systems.
IEEE 2003. Available online: http://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber= 1225051&filter=
AND(p_Publication_Number:8676) (accessed on 15 January 2016).
7. IEC 61727 Ed.2. Photovoltaic (PV) systems—Characteristic of the utility interface. IEC 2004. Available online:
https://webstore.iec.ch/publication/5736 (accessed on 15 January 2016).
8. Bouzid, A.M.; Guerrero, J.M.; Cheriti, A.; Bouhamida, M.; Sicard, P.; Benghanem, M. A survey on control of
electric power distributed generation systems for microgrid applications. Renew. Sustain. Energy Rev. 2015,
44, 751–766. [CrossRef]
9. Teodorescu, R.; Liserre, M.; Rodríguez, P. Grid Converters for Photovoltaic and Wind Power Systems, 1st ed.;
John Wiley and Sons: West Sussex, UK, 2011; pp. 313–321.
10. Blaabjerg, F.; Teodorescu, R.; Liserre, M.; Timbus, A.V. Overview of control and grid synchronization for
distributed power generation systems. IEEE Trans. Indus. Electr. 2006, 53, 1398–1409. [CrossRef]
11. Teodorescu, R.; Blaabjerg, F.; Liserre, M.; Loh, P.C. Proportional-resonant controllers and filters for
grid-connected voltage-source converters. IEE Proc. Electr. Power Appl. 2006, 153, 750–762. [CrossRef]
12. Hu, J.; Shang, L.; He, Y.; Zhu, Z.Q. Direct active and reactive power regulation of grid-connected dc/ac
converters using sliding mode control approach. IEEE Trans. Power Electr. 2011, 26, 210–222. [CrossRef]
13. Fischer, R.J.; González, S.A.; Carugati, I.; Herrán, M.A.; Judewicz, M.G.; Carrica, D.O. Robust predictive
control of grid-tied converters based on direct power control. IEEE Trans. Power Electr. 2013, 29, 5634–5643.
[CrossRef]
14. Chen, D.; Zhang, J.; Zhang, Z. An improved repetitive control scheme for grid-connected inverter with
frequency-adaptive capability. IEEE Trans. Ind. Electr. 2013, 60, 814–823. [CrossRef]
15. Zhao, J.; Bose, B.K. Neural-network-based waveform processing and delayless filtering in power electronics
and AC drives. IEEE Trans. Ind. Electr. 2004, 51, 981–991. [CrossRef]
16. Kang, S.W.; Kim, K.H. Sliding mode harmonic compensation strategy for power quality improvement of a
grid-connected inverter under the distorted grid condition. IET Power Electr. 2015, 8, 1461–1472. [CrossRef]
17. Golestan, S.; Ramezani, M.; Guerrero, J.M.; Freijedo, F.D.; Monfared, M. Moving average filter based
phase-locked loops: Performance analysis and design guidelines. IEEE Trans. Power Electr. 2014, 29,
2750–2763. [CrossRef]
18. Golestan, S.; Guerrero, J.M.; Abusorrah, A.M. MAF-PLL with phase-lead compensator. IEEE Trans. Ind. Electr.
2015, 62, 3691–3695. [CrossRef]
19. Espi, J.M.; Castello, J.; García-Gil, R.; Garcera, G.; Figueres, E. An adaptive robust predictive current control
for three-phase grid-connected inverters. IEEE Trans. Ind. Electr. 2011, 58, 3537–3546. [CrossRef]
20. Svensson, J.; Bongiorno, M.; Sannino, A. Practical implementation of delayed signal cancellation method for
phase-sequence separation. IEEE Trans. Power Del. 2007, 22, 18–26. [CrossRef]
Energies 2016, 9, 190 23 of 23

21. Wang, L.; Jiang, Q.; Hong, L.; Zhang, C.; Wei, Y. A novel phase-locked loop based on frequency detector and
initial phase angle detector. IEEE Trans. Power Electr. 2013, 28, 4538–4549. [CrossRef]
22. TMS320F28335 Digital Signal Controller (DSC)—Data Manual, Texas Instrument, 2008. Available online:
http://www.ti.com/lit/ds/sprs581d/sprs581d.pdf (accessed on 15 January 2016).

© 2016 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons by Attribution
(CC-BY) license (http://creativecommons.org/licenses/by/4.0/).