Professional Documents
Culture Documents
Dynamic Modeling and Closed-Loop Control of Hybrid Grid-Connected Renewable Energy System With Multi-Input Multi-Output Controller
Dynamic Modeling and Closed-Loop Control of Hybrid Grid-Connected Renewable Energy System With Multi-Input Multi-Output Controller
1, January 2021
Abstract—
—In this study, a novel approach for dynamic model‐ greenhouse gas reduction, renewable energy sources such as
ing and closed-loop control of hybrid grid-connected renewable photovoltaic (PV) systems, fuel cells (FCs), and wind tur‐
energy system with multi-input multi-output (MIMO) controller
is proposed. The studied converter includes two parallel DC-DC
bines have been extensively employed to respond to the in‐
boost converters, which are connected into the power grid creasing electrical energy demand. Additionally, the applica‐
through a single-phase H-bridge inverter. The proposed MIMO tion of the PV energy has gained considerable attention in re‐
controller is developed for maximum power point tracking of cent years due to the widespread global existence of solar en‐
photovoltaic (PV)/fuel-cell (FC) input power sources and output ergy. Moreover, with recent developments in solar energy
power control of the grid-connected DC-AC inverter. Consider‐
ing circuit topology of the system, a unique MIMO model is technology, the capital cost of PV systems has been sharply
proposed for the analysis of the entire system. A unique model reduced [2]. It should be noted that the applications of grid-
of the system includes all of the circuit state variables in DC- connected PV systems are preferred due to the removal of
DC and DC-AC converters. In fact, from the viewpoint of storage batteries. In such applications, usually DC-DC and
closed-loop controller design, the hybrid grid-connected energy DC-AC converters are required for the maximum power
system is an MIMO system. The control inputs of the system
are duty cycles of the DC-DC boost converters and the ampli‐ point tracking (MPPT) of the renewable power source and
tude modulation index of DC-AC inverters. Furthermore, the the conversion of the PV output into AC power, respectively.
control outputs are the output power of the PV/FC input power From the viewpoint of controller design, the closed-loop con‐
sources as well as AC power injected into the power grid. After trol of the grid-connected systems is a challenging task due
the development of the unique model for the entire system, a de‐
coupling network is introduced for system input-output linear‐
to the cascade connection of multiple converters.
ization due to inherent connection of the control outputs with Considering a multiport DC-DC converter, a PV panel
all of the system inputs. Considering the decoupled model and and a backup battery are connected to the DC link of a sin‐
small signal linearization, the required linear controllers are de‐ gle-phase grid-connected inverter in [3]. Linear controllers
signed to adjust the outputs. Finally, to evaluate the accuracy are employed for the MPPT of the PV generator and the reg‐
and effectiveness of the designed controllers, the PV/FC based
grid-connected system is simulated using the MATLAB/Simu‐ ulation of the system output power. However, the controllers
link toolbox. are not designed using a dynamic model of the system, and
Index Terms— —Multi-input multi-output (MIMO) converter,
the gains of compensators are selected by trial and error.
maximum power point tracking, grid-connected inverter, con‐ In [4], a bidirectional single-stage grid-connected inverter
version function matrix. is studied for energy storage systems. Several buck-boost
DC-DC choppers have been employed for the connection of
the batteries into the system DC link. Considering the steady-
I. INTRODUCTION state behavior of the system, some equations are obtained
G RID-CONNECTED voltage source inverters have been for power flow control of the storage batteries. The control‐
widely used in active power filters, online uninterrupt‐ ling system does not require any current sensors. As the con‐
able power supplies, and renewable energy systems [1]. In troller is developed based on steady-state analysis, it cannot
these applications, output power control and DC-link voltage guarantee system stability in different operation conditions.
regulation should be performed. Related to the importance of Considering the mathematical model of the multi-parallel
inverters, intrinsic and extrinsic resonances are studied in mi‐
crogrid applications [5]. Moreover, to attenuate the reso‐
Manuscript received: June 17, 2018; accepted: January 23, 2020. Date of
CrossCheck: January 23, 2020. Date of online publication: January 11, 2021. nance of the system, an active damping control approach is
This work was supported by Islamic Azad University–Ardabil Branch. developed. In [5], the PV module and DC-DC converter are
This article is distributed under the terms of the Creative Commons Attribu‐
tion 4.0 International License (http://creativecommons.org/licenses/by/4.0/).
modeled as a constant DC voltage source. However, it is
M. Salimi (corresponding author), F. Radmand, and M. H. Firouz are with the well known that in grid-connected inverters, the DC link has
Department of Electrical Engineering, Ardabil Branch, Islamic Azad University, considerable voltage ripple. Moreover, the developed control‐
Ardabil, Iran (e-mail: m.salimi@iauardabil.ac.ir; f.radmand@iauardabil.ac.ir; hos‐
seini.firooz@iauardabil.ac.ir). ler is unable to track the maximum power point (MPP) of
DOI: 10.35833/MPCE.2018.000353 the input power source.
SALIMI et al.: DYNAMIC MODELING AND CLOSED-LOOP CONTROL OF HYBRID GRID-CONNECTED RENEWABLE ENERGY SYSTEM... 95
In grid-connected PV systems, an attempt is made to maxi‐ studies of the plant. However, dynamic analysis and system
mize the total circuit efficiency by minimizing the number modeling are not studied in [17], [18].
of required converters. For example, in [6]-[8], an attempt is In [19], a novel three-input DC-DC converter for stand-
made to improve the power efficiency of the grid-connected alone PV-FC battery systems is proposed. The MPPT of the
PV systems by removing DC-DC converters. In this condi‐ PV panel and FC system is studied. The linear controllers
tion, the MPPT of the input power source is accomplished are designed based on small signal analysis of the model.
by DC-side voltage control of the inverter. However, in [6]- However, the controller of [19] is only applicable to stand-
[8], series connection of multiple solar panels is necessary to alone renewable systems and dynamic modeling of the DC-
increase DC-link voltage in the transformerless grid-connect‐ AC inverter which is employed in the grid-connected system
ed PV systems. The primary disadvantage of the panels’ se‐ is not reported. It is clear that adding the AC-side state vari‐
ries connection appears during partial shading. Due to the ables into the converter model complicates the development
current reduction in the shaded cells, the total output power of the unique general model for the entire system, and also
will decrease considerably in series-connected PV panels. increases the difficulty of the controller design.
Conversely, the application of 50 Hz bulky transformers in In [20], the closed-loop control of the grid-connected in‐
the inverter output for the grid connection of the PV systems verters is studied to improve the current quality and active
is undesirable due to its high weight and cost. Briefly, a volt‐ power filtering in the distribution system. In spite of the non-
age boost of the PV output power is required in grid-con‐ ideal waveform of the grid voltage, the developed controller
nected systems by using a step-up DC-DC converter [9]-[11]. is able to compensate nonlinear and harmonic components
Low-frequency power mitigation control is studied for of the local load. Furthermore, it is not required to transfer
grid-connected PV systems by using a dynamic model based coordinates of system voltages and currents. Hence, the con‐
controller [12]. Considering low-voltage ripple of the DC trol law can be calculated simply, which results in ease of
link, small capacitors can be used in the grid-connected in‐ practical implementation.
verter. In addition, the MPPT of input PV sources can be In [21], the nonlinear robust control of PV systems is stud‐
performed with fast tracking speed. However, a developed ied. The proposed controller is designed to track the MPP of
controller cannot be employed in hybrid renewable energy the input power source, regulate DC voltage of the inverter,
systems. Due to the dependence of the generated PV power as well as active power injected into the power grid. To pro‐
on radiance level, temperature, and shading conditions, and tect the system from inverter faults during voltage dips, a re‐
the lack of power generation during night hours, PV systems active power injection capability is provided for the system.
are normally combined with other energy sources or storage It can be used during normal grid-connected conditions as
systems to improve the reliability of the renewable power well as stand-alone fault mode. Two separate models at AC
plant. If the hybrid renewable energy system is equipped inverter side and DC chopper side are employed for nonlin‐
with an appropriate controller, it can demonstrate better reli‐ ear control of the system.
ability for load power supply [13] compared with a single- Utility fault ride-through capability of the grid-connected
source system. FC generator is an ideal option for combina‐ renewable energy systems is investigated in [22] using pre‐
tion with PV systems [14] due to acceptable power efficien‐ dictive controllers. In addition to standard capabilities of the
cy and independence from weather conditions. grid-connected PV system, the controller can provide reac‐
Recently, the research and development of the hybrid re‐ tive power for load compensation and unity power factor op‐
newable energy system have increased. For example, in [15], eration. During voltage dips, the controller stops active pow‐
a backstepping nonlinear controller is designed for hybrid er injection and the system operates as an active power filter.
PV power supplies for remote communication applications. In [23], the circuit topology of current source grid-con‐
To supply system loads, a buck/boost DC-DC converter is nected inverters is modified to improve system response dur‐
cascaded with a single-phase DC-AC inverter. A nonlinear ing voltage spikes of the grid voltage. It is realized by add‐
backstepping controller guarantees the stability and robust‐ ing a standard buck chopper between input renewable power
ness of the designed controllers during system parameter source and a current source inverter. Hence, in this structure,
changes. However, in [15], the DC-DC converter and DC- the PV panel current pulsates resulting in the oscillation of
AC inverter are modeled as separate subsystems. Hence, in‐ the operation point.
herent connections between these subsystems are not consid‐ Briefly, in grid-connected renewable systems, a DC-DC
ered. In fact, the multi-input multi-output (MIMO) model of converter can be cascaded with a DC-AC inverter for the
the cascaded converters is not used for controller design. grid connection of the input power source. In this structure,
Moreover, the practical implementation of the backstepping a DC-DC converter is employed for MPPT of the renewable
controller [16] requires very fast processors and digital/ana‐ power source. Additionally, a DC-AC inverter is responsible
log (D/A) converters. Furthermore, gains of the controllers for AC power flow control between the renewable generator
in [15] are selected by trial and error. and the utility. To simplify the controller design, the input
The analysis of a grid-connected PV-wind-battery hybrid power source and the DC-DC chopper can be approximated
energy system is presented in [17], [18], which includes en‐ as a constant DC voltage source [12]. In this condition, the
ergy engineering concepts, reliability, and cost reduction modulation index (control input) of the grid-connected invert‐
96 JOURNAL OF MODERN POWER SYSTEMS AND CLEAN ENERGY, VOL. 9, NO. 1, January 2021
er is employed for AC current (control output) control. II. CONVERTER TOPOLOGY AND MODELING
Hence, the renewable energy system can be approximated as The topology of the MIMO converter, which will be con‐
a single-input single-output (SISO) system. sidered in this study, is shown in Fig. 1. According to bipo‐
However, the DC link of the grid-connected inverter may lar pulse width modulation of the inverter [25], it can be
include considerable voltage ripples. Hence, it cannot be ap‐ concluded that the inverter output voltage is ux 3. Moreover,
proximated by a simple constant voltage source in a wide its input current will be equal to ux 4. u, x 3, and x 4 are the
range of operation. In brief, to provide more accurate system amplitude modulation index, inverter DC-link voltage, and
control, the cascaded DC-DC and DC-AC converters must grid current, respectively. Considering these assumptions, the
be considered as a unique system in grid-connected renew‐ equivalent circuit of the system is shown in Fig. 2. The pro‐
able energy systems. In this condition, the duty cycle of the posed converter has two input sources v 1 (PV) and v 2 (FC).
chopper should be considered for the MPPT of the input The DC-DC converters are employed for the MPPT of the
power source. Furthermore, the modulation index of the in‐ input power sources. Duty cycles of the switches S 1 and S 2
verter is employed for output power control. In fact, if a (d 1 and d 2) are control inputs of the DC-DC boost convert‐
unique model is employed for controller design, the grid-con‐ ers. Additionally, u is the control input of the DC-AC grid-
nected system will include two different control inputs (in‐ connected inverter in the hybrid energy system. Equivalent
cluding the duty cycle of the chopper and the modulation in‐ circuits of the MIMO converter in different operation modes
dex of the inverter) and two different control outputs (includ‐ are shown in Fig. 3, where T is the switching period of DC-
ing the input power of the renewable energy source and the DC boost converters. Moreover, the steady-state waveforms
grid current). Hence, the mentioned renewable generator is a of the DC-DC boost converters are illustrated in Fig. 4,
MIMO system [24]. In spite of the superior response of the where X G1 and X G2 are the gate signals of the S 1 and S 2, re‐
MIMO design, considering inherent connection among con‐ spectively.
trol inputs and outputs, it is clear that the MIMO controller x1 L1 D1
design is significantly challenging compared with SISO sys‐ d 1
PV source S1
tems.
In this study, the modeling and control of the grid-connect‐ x2 L2 D2 ux4
ed PV/FC hybrid energy systems are studied. The circuit to‐
+ S3 S5 x4 Lg
pology consists of two separate DC-DC boost converters for d2
x3 C +
FC source S2
the MPPT of the input sources. These choppers supply the
S4 S6
ux
3
vg
DC link of the grid-connected inverter. In the proposed ap‐
proach, a complete unique MIMO model of the system is
Fig. 1. Grid-connected PV/FC hybrid energy system.
employed for controller design. The control inputs of the sys‐
tem are duty cycles of the DC-DC boost converters and the x1 L1 D1
amplitude modulation index of the DC-AC inverter. In addi‐
tion, the system control outputs are the power from the PV/ v1 + S1
FC sources and the AC power injected into the power grid.
Hence, from the viewpoint of controller design, the studied x2 L2 D2
system is a three-input three-output circuit. Considering the Lg x4
complexities of controller design due to inherent connection +
v2 + S2 x3 C ux ux3 + vg
between control outputs and all of the system inputs, a spe‐
4
cial decoupling network is employed to decouple controlling
loops. Finally, considering small signal linearization of the Fig. 2. Equivalent circuit of grid-connected PV/FC hybrid energy system.
MIMO model, system controllers are designed according to
Bode analysis. Briefly, the primary contributions of this According to Fig. 4, three different operation modes can
study can be summarized as follows: be considered as 0 < t < d 1 T, d 1 T < t < d 2 T, and d 2 T < t < T in‐
1) Develope a general unique model for the MIMO hybrid tervals. By choosing the system state vector as X T =
energy system. [i L i L v C i L ]=[x 1 x 2 x 3 x 4 ], an averaged state-space model
1 2 g
2) Develope a decoupling network and systematic design
of the system can be extracted, where v C is the output capaci‐
of the linear controllers for decoupled loops using Bode anal‐
tor voltage; i L is the PV current; i L is the FC current; and
ysis. 1 2
The rest of this paper is organized as follows. The convert‐ i L is the grid current. During the first operation mode for 0 <
g
er structure and its analysis are explained in Section II. t < d 1 T, it is observed that switches S 1 and S 2 are on. Hence,
Then, dynamic modeling of the converter is introduced in inductors L 1 and L 2 are charged through input voltage sourc‐
Section III. In Section IV, controller design and decoupling es v 1 and v 2, respectively. Considering Fig. 3(a), state-space
networks are presented. Finally, the simulation results of the equations of the related sub-circuit can be written as follows:
controller are illustrated in Section V. Ẋ = A 1 X + B 1 (1)
SALIMI et al.: DYNAMIC MODELING AND CLOSED-LOOP CONTROL OF HYBRID GRID-CONNECTED RENEWABLE ENERGY SYSTEM... 97
x1 L1 D1 ì é0 0 0 0 ù
ï ê0 0
ïA = ê 0 0 úú
v1 + S1
ï 1 ê0 0 0 - u Cú
ï ê ú
ï ë0 0 0 0 û
x2 L2 D2 ï
Lg x4 í éV 1 L 1 0 0 0 ù (2)
ï ê ú
+
x3 C ux ï êê 0 V2 L2 0 0 úú
v2 + S2 4 ux3 + vg ïB = êê úú
ï 1 êê 0 0 0 0 úú
ï
ï ê 0 0 0 - V g L g úû
(a) î ë
x1 L1 D1
where C is the DC-link capacitor of the inverter; V 1 and V 2
v1 + S1 are the voltages of PV and FC sources, respectively; V g is
the grid voltage; and L g is the coupling inductor.
x2 L2 D2 According to Fig. 3(b), if it is assumed that d 1 < d 2, it can
Lg x4 be concluded that S 1 will be turned off at t = d 1 T during the
+
v2 + S2 x3 C ux ux3 + vg second operation interval. However, S 2 is still conducting.
4
Hence, the voltage of L 1 will be equal to v 1 - v C. Similarly,
the following equations can be written during the second
(b) switching interval:
x1 L1 D1
Ẋ = A 2 X + B 2 (3)
v1 + S1
ì é 0 0 - 1 L1 0 ù
ï ê ú
x2 L2 D2 ï ê 0 0 0 0 ú
Lg x4 ïïA = ê ú
+ ïï 2 ê1 C 0 0 - u Cú
ïï
v2 + S2 x3 C ux
4 ux3 + vg ï ê 0 0 u Lg 0 úû
ï ë
í (4)
ï éV 1 L 1 0 0 0 ù
(c) ï ê ú
ïï êê 0 V2 L2 0 0 úú
Fig. 3. Equivalent sub-circuits of converter in different operation modes. ïïB 2 = êê úú
(a) The first kind of equivalent sub-circuit during 0 < t < d 1 T. (b) The second ïï êê 0 0 0 0 úú
ï
kind of equivalent sub-circuit during d 1 T < t < d 2 T. (c) The third kind of ï ê 0 0 0 - V g L g úû
equivalent sub-circuit during d 2 T < t < T. î ë
Finally, in the third switching interval (d 2 T < t < T), it is
Carrier signal
observed that both of the switches S 1 and S 2 are off and the
d1
state equations of the sub-circuit can be obtained according
d2
to Fig. 3(c).
XG1
t Ẋ = A 3 X + B 3 (5)
ì é 0 0 - 1 L1 0 ù
ï ê ú
t ï êê 0 0 - 1 L2 0 úú
XG2 ïA = êê úú
ï 3 êê1 C 1 C 0 - u C úú
ï
ï ê 0 0 u Lg 0 úû
t ï ë
iL1 í (6)
ï éV 1 L 1 0 0 0 ù
ï ê ú
ï êê 0 V2 L2 0 0 úú
t ïB 3 = êê úú
iL2 ï êê 0 0 0 0 úú
ï
ï ê 0 0 0 - V g L g úû
î ë
d1 T (d2d1)T (1d2)T t
d2T
III. DYNAMIC MODELING OF CONVERTER
T
Fig. 4. Steady-state waveforms of DC-DC boost converters in grid-con‐ A. Averaged State Space MIMO Model
nected PV/FC hybrid energy system. In the hybrid converter, duty cycles (d 1, d 2) of the boost
98 JOURNAL OF MODERN POWER SYSTEMS AND CLEAN ENERGY, VOL. 9, NO. 1, January 2021
tained as:
ï v͂ =[d͂ 1 d͂ 2 u͂ ]
ï
é d1 - 1 ù ï y͂ =[i͂ L i͂ L i͂ L ]
ê 0 0 0 ú é V1 ù î 1 2 g
ê L1 ú ê 0 0 0 ú
ê ú ê L1 ú where i͂ L , i͂ L , v͂ C, and i͂ L are the small signal perturbations of
d2 - 1 êê úú
éẋ 1 ù ê 0 0 úú éê 1 ùú
x V2
1 2 g
0 êê 0
êẋ ú ê
ê 2ú = ê L2 ú êêx 2 úú + êê 0 0 úúú the i L , i L , v C, and i L around an operation point, respectively.
L2 ú 1 2 g
êẋ 3 ú ê 1- d ê ú êê úú
1- d 2 u ú êx 3 ú êê 0
ê ú ê
ëẋ 4 û ê C
1
0 - úú êëx úû êê 0 0 0 úúú IV. CONTROLLER DESIGN
C C 4 ú
ê ú ê Vg ú
ê u ú ê0 0 0 - ú The block diagram of the proposed controller for a grid-
ê 0 0 0 ú ë Lg û connected PV/FC hybrid energy system is shown in Fig. 5,
ë Lg û
where SPWM stands for sinusoidal pulse width modulation.
(10) The duty cycles of the DC-DC converters and inverter ampli‐
tude modulation index are the control inputs of the system.
B. Model Linearization Furthermore, the output power of the PV/FC sources and in‐
Considering small-signal linearization theory, state vari‐ jected AC current are the control outputs.
ables, duty cycles, and amplitude modulation, the index of
x1 L1 D1
the model can be assumed as:
PV
ìx i = Xˉi + x͂ i i = 1234
S1
source
ï
íd j = Dˉ j + d͂ j j = 12 (11) x2 L2 D2
ï
îu = Uˉ + u͂ + S3 S5 Lg
FC x3 x4
where DC components (Xˉi Dˉ j Uˉ) are the nominal values of
S2 vg
source
S4 S6
the parameters in an operation point of the converter; and
AC components (x͂ i d͂ j u͂ ) are the small signal perturbations Controller MPPT MIMO controller
Controller 1
of the parameters around an operation point. KPV x1
x1ref K1(1+T1s) d1
By substituting (11) into (10), the linearized small-signal ISC,PV ×
+
+ PWM S1
s(1+aT1s)
model of the converter can be obtained. It should be noted
Controller 2
that in the linearized model, the higher-order terms where KFC x2
x2ref K2(1+T2s) d2
two AC components are multiplied, are neglected. ISC,FC ×
+
+ PWM S2
{
s(1+aT2s)
ẋ͂ = Ax͂ + Bv͂
(12)
y͂ = Cx͂ x3 PI controller x4
Controller 3
x3ref K K4(1+T4s) u Bipolar
where x͂ , v͂ , and y͂ are the state variable vector, control vec‐ +
+ K p+ s i ×
+
+
s(1+aT4s) SPWM
S3
of the linearized state space model can be rewritten as: Fig. 5. Proposed control structure for grid-connected PV/FC hybrid energy
system.
SALIMI et al.: DYNAMIC MODELING AND CLOSED-LOOP CONTROL OF HYBRID GRID-CONNECTED RENEWABLE ENERGY SYSTEM... 99
As shown in Fig. 5, the first control input d 1 is adjusted From (16), it can be observed that all of the output vari‐
so that the operation of the PV source in MPP is established. ables of the system (x͂ 1 , x͂ 2, and x͂ 4) are related to all of the
To do this, at first, the reference value of the PV array cur‐ control inputs d͂ 1, d͂ 2, and u͂ . Hence, it is not straightforward
rent x ref
1 , which guarantees its operation in MPP, is calculat‐ to design MPPT and AC current controllers. For this reason,
ed. Then, as shown in Fig. 5, the first controller determines a compensation network [27] is employed in this study to fa‐
d 1 according to the error of the first state variable (e 1 = x ref
1 - cilitate the design process of closed-loop controllers, as
x 1). Similarly, d 2 is used for the MPPT of the FC power shown in Fig. 6.
source. The values of the reference currents for PV and FC
input power sources can be calculated according to the frac‐ v G* v* G y
tional short-circuit approach as x ref i = K PV I SCPV and x i =
ref
K1(1+T1s) d1 1 g11
x1ref + + x1
K FC I SCFC, where K PV = 0.75 and K FC = 0.85 are the coeffi‐ s(1+aT1s) g21
* g21
cients of the PV and FC reference currents, respectively; and x1 g31
* g31
I SCPV and I SCFC are the short-circuit currents of the PV and K2(1+T2s) d2
g12
*
g22
*
g12
g22
FC, respectively [26]. x2ref + + x2
s(1+aT2s)
g32
* g32
Alternatively, the amplitude modulation index of the in‐ x2
g13
* g13
verter u is controlled to adjust the AC grid current on its ref‐ g23
* g23
K4(1+T4s) u
erence value. It is obvious that the grid reference current x4ref + + x4
s(1+aT4s) g33
* g33
must be selected so that all of the power generated by renew‐
x4
able input sources is injected into the power grid. To accom‐
plish this, another separate controller is employed for the Fig. 6. Compensation network.
voltage control of the DC-link capacitor.
If the voltage of the DC-link capacitor is stabilized at a In fact, the compensation network is capable of decou‐
certain reference value x ref3 , it can be concluded that all of pling multiple control loops of the proposed strategy. In the
the active power generated by DC sources is transferred into mentioned compensation network, the output vector y can be
the power grid. In fact, if the power generated by renewable defined as y = Gv *, where v * is the modified control inputs,
sources from one side and the injected AC power into the which can be defined as v * = G * v. Therefore, it can be con‐
power grid from the other side are not equal, the power dif‐ cluded that y = GG * v. Thus, to decouple the control loops,
ference will be transferred into the DC-link capacitor and the GG * must be a diagonal unity matrix. Then, each control
therefore, the voltage of the DC-link capacitor cannot be output will only be related to the corresponding control in‐
kept constant at its reference value. Briefly, the stabilization put. With respect to these equations, it can be concluded that
of the DC-link voltage at its reference value and the calcula‐ G * = G -1 xv -1, which is given in (17).
tion of the power grid reference current based on the DC- g 13 g 32 - g 12 g 33 g 12 g 23 - g 13 g 22 ù
é
link voltage error ensure that all of the power generated by ê 1 ú
renewable sources is transferred into the power grid. ê g 11 g 33 - g 13 g 31 g 11 g 22 - g 12 g 21 ú
êê úú
As shown in Fig. 5, a cascade structure is proposed for êê g 23 g 31 - g 21 g 33 g 13 g 21 - g 11 g 23 úú
G = êê
*
1 ú (17)
closed-loop control of the grid-connected inverter. At first, êê g 22 g 33 - g 23 g 32 g 11 g 22 - g 12 g 21 úúú
according to the DC-link voltage error, the proportional-inte‐ êê úú
gral (PI) controller determines the amplitude of the grid ref‐ ê g 21 g 32 - g 22 g 31 g 12 g 31 - g 11 g 32 ú
êg g - g g 1 ú
erence current. Then, another controller is employed for reg‐ ë 22 33 23 32 g 11 g 33 - g 13 g 31 û
ulation of the grid current. It is clear that for satisfactory According to Fig. 6, by substituting G * and G from (17)
control of the AC current controller, the DC-link reference and (16) into x = GG * v, the decoupled transfer functions of
voltage must be larger than the peak value of the grid volt‐ the system can be obtained as:
age. It should be noted that in this study, the reference value
x͂ 1 g 23 g 31 - g 21 g 33 g 21 g 32 - g 22 g 31
of the DC-link capacitor is assumed to be 200 V. H1 (s) = = g 11 + g 12 + g 13
Considering the Laplace transform of (12), the following d͂ 1 g 22 g 33 - g 23 g 32 g 11 g 22 - g 12 g 21
equations can be obtained: (18)
y͂ = Gv͂ (14) x͂ 2 g 13 g 32 - g 12 g 33 g 12 g 31 - g 11 g 32
H2 (s) = = g 21 + g 22 + g 23
G = C(sI - A) -1 B (15) ͂d 2 g 11 g 33 - g 13 g 31 g 11 g 33 - g 13 g 31
where G is the transfer function of the system; s is the oper‐ (19)
ator variable in Laplace domain; and I is an identity matrix. x͂ 4 g 12 g 23 - g 13 g 22 g 13 g 21 - g 11 g 23
Considering the inputs and outputs of the proposed MIMO H4 (s) = = g 31 + g 32 + g 33
system, (14) can be rewritten as follows: u͂ g 11 g 22 - g 12 g 21 g 11 g 22 - g 12 g 21
(20)
éy 1 ù éx͂ 1 ù ég 11 g 12 g 13 ù éd͂ 1 ù
ê ú ê ú ê úê ú Considering the developed transfer functions in (18)-(20),
êy 2 ú = êx͂ 2 ú = êg 21 g 22 g 23 ú êêd͂ 2 úú (16) the proposed control loops in Fig. 6 can be designed. It
ê ú ê úê ú
y
ë 3û ë x͂ 4 û g
ë 31 g 32 g 33 û ë u û
͂ should be noted that the control functions in the Laplace do‐
The simplified values of the transfer functions (gij, i, j = main are assumed as K i (1+ T i s) [s(1+ aT i s)], where i = 124.
1, 2, 3) are presented in the Supplementary Material. Therefore, it is possible to eliminate the steady-state error of
100 JOURNAL OF MODERN POWER SYSTEMS AND CLEAN ENERGY, VOL. 9, NO. 1, January 2021
-180
+
-270
Iph vPV
-360
100 101 102 103 104 105 106
( (v
id=IO exp PV +
Vta
RsiPV
) 1 )
Frequency (rad/s)
(b) Fig. 8. Equivalent circuit of PV input source.
100
Magnitude (dB)
TABLE III
50 NOMINAL PARAMETERS OF POWER CIRCUIT
0
-50 Simulation parameter Value
-100 Switching frequency of inverter and DC-DC
20 kHz
180 converters f s
Phase (°)
90 PV inductor L 1 1 mH
0
FC inductor L 2 1 mH
-90
-180 Grid inductor L g 1 mH
100 101 102 103 104 105 106 DC-link capacitor C 470 μF
Frequency (rad/s) Voltage of PV sources V 1 159.3 V
(c)
Voltage of FC sources V 2 150 V
After compensation; Before compensation
AC grid frequency f 50 Hz
Fig. 7. Frequency response of closed-loop system before and after compen‐
sation. (a) Bode plot of H 1 ( s ) before and after compensation. (b) Bode plot AC grid voltage V g 110 V
of H 2 ( s ) before and after compensation. (c) Bode plot of H 4 ( s ) before and
after compensation.
Hence, current-voltage characteristics of the PV panel can
TABLE I be written as:
( ( ) )
CONTROLLER GAINS
v PV + R s i PV
i PV = I ph - I O exp -1 (21)
Gain V ta
Controller
Ki Ti aT i where i PV and v PV are the current and voltage of the PV pan‐
Current controller of PV el, respectively; I ph and I O are the photo-generated and reverse
source H 1 ( s ) 0.98550 0.007491 0.00001916
saturation currents, respectively; R s is the series resistance;
Current controller of FC
0.32570 0.005260 0.00002005
and V ta = N S AK B T PV q is the thermal voltage, and A, K B, q,
source H 2 ( s )
and T PV are the diode quality factor, Boltzmann constant, elec‐
AC current controller of tron charge, and PV module temperature, respectively.
power grid H 4 ( s ) 0.15835 0.014893 0.00000562
In addition, the photo-generated current I ph depends on the
SALIMI et al.: DYNAMIC MODELING AND CLOSED-LOOP CONTROL OF HYBRID GRID-CONNECTED RENEWABLE ENERGY SYSTEM... 101
irradiance level and temperature as follows: with zero steady-state error. Moreover, the THD of the grid
S current is shown in Fig. 10(f), where when the fundamental
I ph = I SC + C t (T a - T 0 ) (22) (50 Hz) is equal to 22.55, THD equals 1.48%. The THD of
S0
the injected AC current is equal to 1.47%, which is com‐
where I SC is the short-circuit current of the panel; T 0 and S 0 pletely compatible with the standards of a distribution net‐
are the nominal temperature and irradiance, respectively; S work.
and T a are the ambient irradiance and temperature, respec‐ In Fig. 11, the transient response of the developed MIMO
tively; and C t is the temperature coefficient. controller is illustrated during system start-up process. It is
Furthermore, as shown in Fig. 9, a controlled voltage assumed that reference values of the input sources are
source E with an internal resistance R can be considered for changed from zero to nominal values. Furthermore, consider‐
modeling the FC stack in nominal conditions. During the ing step changes of the reference voltage of DC-link capaci‐
simulations, it is assumed that E = 150 V and R = 0.2 Ω. tor from 0 to 200 V, the grid reference current is determined
in the outer loop of the AC-side MIMO controller. It is ob‐
R
+
served that the proposed MIMO controller is stable and fast
E ifc
+
vfc during system start-up process.
Current (A)
Current (A)
10 Reference; IL 1 10 Reference; I L2
Fig. 9. Equivalent circuit of FC input source.
5 5
0 0
In Fig. 10, the simulation response of the proposed MI‐ 0.40 0.42 0.44 0.46 0.48 0.50 0.40 0.42 0.44 0.46 0.48 0.50
Time (s) Time (s)
MO controller for the grid-connected PV/FC hybrid energy (a) (b)
system in nominal condition is illustrated during steady-state
operation. Reference; VC Reference; IL g
Voltage (V)
Current (A)
300 40
200
0
Current (A)
Current (A)
Current (A)
200 0.8
from 1000 to 600 W/m 2 at t = 0.5 s. Hence, the reference cur‐
Vg (V)
100 0.6
-100 0.4
-200
0.2 rent of the PV panel is changed from 4.7 to 2.8 A. The pro‐
0.40 0.42 0.44 0.46 0.48 0.50 0 2 4 6 posed MIMO controller is able to regulate input current of
Time (s) Frequency (kHz)
(e) (f) the PV panel in MPP. Moreover, in spite of grid current
changes, the DC-link voltage is stably regulated in its refer‐
Fig. 10. Steady-state response of proposed MIMO controller for grid-con‐
nected PV/FC hybrid energy system. (a) Output current of PV source. (b)
ence value.
Output current of FC stack. (c) Voltage of DC-link capacitor. (d) Grid cur‐
Reference; IL1 Reference; IL 2
Current (A)
Current (A)
rent versus its reference value. (e) Grid voltage. (f) Total harmonic distor‐ 10 10
tion (THD) of grid current. 5 5
0 0
0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.4 0.5 0.6 0.7 0.8 0.9 1.0
It is assumed that T a = 25 ℃ and S = 1000 W/m . Hence, 2 Time (s) Time (s)
(a) (b)
the reference current of the PV panel will be equal to 4.7 A
Voltage (V)
Current (A)
10 Reference; IL 1 20 Reference; IL 2
5 10 well as the injection of generated power into the power grid.
0 0 Considering the dependence of the control outputs on all of
0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.4 0.5 0.6 0.7 0.8 0.9 1.0 the control inputs, a special compensation network is devel‐
Time (s) Time (s)
(a) (b) oped to decouple the control loops. According to the frequen‐
cy response of the system transfer functions, the control
Voltage (V)
300 Reference; VC gains of the developed MIMO controller are tuned. The accu‐
200
racy and effectiveness of the designed MIMO controller are
100
0.4 0.5 0.6 0.7 0.8 0.9 1.0 evaluated by some simulations in the MATLAB/Simulink
Time (s) toolbox. According to the simulation results, the proposed
(c) MIMO controller is completely fast and stable at different
Fig. 13. Transient response of proposed MIMO controller during step operation points with zero steady-state error. Moreover, the
changes of FC reference current. (a) Current of PV panel. (b) Current of FC THD of the grid current is approximately 1.48%, which is
stack. (c) DC-link voltage.
compatible with the standards of distribution networks.
Current (A)
Current (A)
Current (A)
300 60
200 penetration of renewables: an overview of interlinked technical, envi‐
100 0 ronmental and socio-economic challenges,” Journal of Modern Power
0 -60 Systems and Clean Energy, vol. 7, no. 1, pp. 1-8, Aug. 2018.
0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.4 0.5 0.6 0.7 0.8 0.9 1.0 [3] A. Ajami and P. A. Shayan, “Soft switching method for multiport DC/
Time (s) Time (s) DC converters applicable in grid connected clean energy sources,” IET
(c) (d) Power Electronics, vol. 8, no. 7, pp. 1246-1254, Jul. 2015.
200 [4] K. Y. Lo, Y. M. Chen, and Y. R. Chang, “Bidirectional single-stage
Vg (V)
Current (A)
maximum power point tracking,” IEEE Transactions on Power Elec‐ [24] J. J. E. Slotine and W. Li, Applied Nonlinear Control. London: Pren‐
tronics, vol. 31, no. 1, pp. 328-339, Jan. 2016. tice Hall Press, 1991.
[13] J. Hu, Y. Shan, Y. Xu et al., “A coordinated control of hybrid AC/DC [25] N. Mohan, T. M. Undeland, and W. P. Robbins, Power Electronics:
micro-grids with PV-wind-battery under variable generation and load Converters, Applications and Design. New York: John Wiley & Sons
conditions,” International Journal of Electrical Power & Energy Sys‐ Press, 1995.
tems, vol. 104, pp. 583-592, Jan. 2019. [26] H. A. Sher, A. F. Murtaza, K. E. Addoweesh et al., “A new sensorless
[14] A. Djafour, M. S. Aida, and B. Azoui, “Photovoltaic assisted fuel cell hybrid MPPT algorithm based on fractional short-circuit current mea‐
power systems,” Energy Procedia, vol. 50, pp. 306-313, Jul. 2014. surement and P&O MPPT,” IEEE Transactions on Sustainable Energy,
[15] A. D. Martin, J. M. Cano, J. F. A. Silva et al., “Backstepping control vol. 6, no. 5, pp. 1426-1434, Jul. 2015.
of smart grid-connected distributed photovoltaic power supplies for [27] Z. Qian, O. Abdel-Rahman, M. Pepper et al., “Analysis and design for
telecom equipment,” IEEE Transactions on Energy Conversion, vol. paralleled three-port DC/DC converters with democratic current shar‐
30, no. 4, pp. 1496-1504, Dec. 2015. ing control,” in Proceedings of 2009 IEEE Energy Conversion Con‐
[16] M. Salimi, J. Soltani, G. A. Markadeh et al., “Indirect output voltage gress and Exposition, San Jose, USA, Sept. 2009, pp. 1375-1382.
regulation of DC-DC buck/boost converter operating in continuous
and discontinuous conduction modes using adaptive backstepping ap‐
proach,” IET Power Electronics, vol. 6, no. 4, pp. 732-741, Apr. 2013. Mahdi Salimi received the B.S. and M.S. degrees in electrical engineering
[17] C. Peng, P. Xie, L. Pan et al., “Flexible robust optimization dispatch from the K. N. Toosi University of Technology, Tehran, Iran, in 2000 and
for hybrid wind/photovoltaic/hydro/thermal power system,” IEEE 2002, respectively, and the Ph.D. degree in power electronics from Science
Transactions on Smart Grid, vol. 7, no. 2, pp. 751-762, Mar. 2016. and Research Branch, Islamic Azad University, Tehran, Iran, in 2012. Since
[18] H. M. Al-Masri and M. Ehsani, “Feasibility investigation of a hybrid 2003, he has been with the Islamic Azad University, where he is presently
on-grid wind photovoltaic retrofitting system,” IEEE Transactions on working as an Assistant Professor in the Department of Electrical Engineer‐
Industry Applications, vol. 52, no. 3, pp. 1979-1988, May-Jun. 2016. ing. His research interests include the closed-loop control of power electron‐
[19] J. D. Dasika, B. Bahrani, M. Saeedifard et al., “Multivariable control ics converters, high-gain DC-DC converter, grid-connected inverter and re‐
of single-inductor dual-output buck converters,” IEEE Transactions on newable energy.
Power Electronics, vol. 29, no. 4, pp. 2061-2070, Apr. 2014.
[20] X. Guo and J. M. Guerrero, “Abc-frame complex-coefficient filter and
Fereshteh Radmand received the B.Sc. degree in electrical power engineer‐
controller based current harmonic elimination strategy for three-phase
ing from the Islamic Azad University of Ahar, Ahar, Iran, in 2013, and her
grid connected inverter,” Journal of Modern Power Systems and
M. Sc. degree in electrical power engineering from the University of Ahar,
Clean Energy, vol. 4, no. 1, pp. 87-93, Jan. 2016.
Ahar, Iran, in 2015, where she is presently working towards the Ph.D. de‐
[21] A. Merabet, L. Labib, A. M. Y. M. Ghias et al., “Robust feedback lin‐
gree in electrical engineering. Her current research interests include power
earizing control with sliding mode compensation for a grid-connected
electronic converter and control, and renewable energy system.
photovoltaic inverter system under unbalanced grid voltages,” IEEE
Journal of Photovoltaics, vol. 7, no. 3, pp. 828-838, May 2017.
Mansour Hosseini-Firouz received the B. E. degree in electrical engineering
[22] A. Merabet, L. Labib, and A. M. Y. M. Ghias, “Robust model predic‐
from the Ardabil Branch, Islamic Azad University, Ardabil, Iran, in 2003, the
tive control for photovoltaic inverter system with grid fault ride-
M.S. and Ph.D. degrees in electrical engineering from the Science and Research
through capability,” IEEE Transactions on Smart Grid, vol. 9, no. 6,
Branch, Islamic Azad University, Tehran, Iran, in 2005 and 2013, respectively.
pp. 5699-5709, Nov. 2018.
He is currently an Assistance Professor at the Ardabil Branch, Islamic Azad Uni‐
[23] Y. Geng, K. Yang, Z. Lai et al., “A novel low voltage ride through
versity. His research interests include power system, power electronics, reliabili‐
control method for current source grid-connected photovoltaic invert‐ ty and electricity market.
ers,” IEEE Access, vol. 7, pp. 51735-51748, Apr. 2019.