Version of Record: https://www.sciencedirect.

com/science/article/pii/S0038092X20304941
Manuscript_3ee45323118d40d32197062f1d25d767

1 Grid Integration of a PV System Supporting an EV Charging Station Using

2 Salp Swarm Optimization
3 Ahmed A. S. Mohameda, b, Senior Member, IEEE, Ahmed El-Sayedc, Hamid Metwallya and Sameh
4 I. Selema
a
5 Electrical Power and Machines Department, Faculty of Engineering, Zagazig university, Zagazig, Egypt
b
6 National Renewable Energy Laboratory (NREL), Golden, CO, USA
c
7 Engineering Management, Zagazig university, Zagazig, Egypt
8

9 Abstract

10 Integrating photovoltaic (PV)-based power systems with electric vehicle (EV) charging loads
11 makes perfect sense due to the similarities between them in terms of power form, interface, and
12 locations. However, this integration involves a high level of uncertainty, due to the random
13 nature of solar generation and EV charging, which requires a careful control design and
14 optimization to keep the system stable. This paper presents an optimization algorithm to find the
15 best combination of the control parameters of a voltage source inverter that integrates a PV
16 power system with an EV charging station through a common grid-connected ac-bus. The
17 controller parameters are optimized using Salp Swarm Algorithm to minimize the fluctuation in
18 the dc-bus voltage through balancing the active power-flow and the injected harmonics to the
19 grid. The controller performance under different severe disturbances from both generation and
20 load sides is investigated and compared with an optimum control design using system transfer
21 functions and particle swarm optimization. A hypothetical level 2 ac charging station for EVs is
22 modeled considering different operating conditions and utilized to test the controller along with
23 real-world irradiance profiles. The proposed controller is validated using simulation and
24 processor-in-the-loop platforms. The outcomes show that the proposed control design is able to
25 reduce the fluctuation of the dc-bus voltage by around 50%, the total harmonic distortion of
26 voltage by 40% and current by 64% compared to the analytical-based design, which makes the
27 system compatible with the requirements defined by the IEEE 1547 international standard.

28 Keywords: Photovoltaic (PV) system, Salp Swarm Algorithm (SSA), Voltage Source Inverter (VSI),
29 PI control, Electric Vehicle (EV), Optimal Control.
30

1|Page
© 2020 published by Elsevier. This manuscript is made available under the Elsevier user license
https://www.elsevier.com/open-access/userlicense/1.0/
31

32 1. Introduction
33 With the large-scale deployment of electric vehicles (EVs), the power grid is expected to
34 experience a major challenge to satisfy the load demand while keeping the system stable. The
35 increasing number of EVs will add additional stresses to the existing distribution system’s
36 components, such as transformers and cables, and may perturb their operation, particularly
37 during the peak demand periods (Karfopoulos et al., 2014; Li et al., 2016; Xiao et al., 2014, p. 0;
38 Yilmaz & Krein, 2013). These impacts can be minimized if EVs charging are integrated with
39 renewable energy sources (RESs), such as solar and wind generations (Moeini-Aghtaie et al., 2017;
40 Quevedo et al., 2017; Ugirumurera & Haas, 2017). The RES can be designed to supply the bulk
41 power demand of EVs charging with minor support from the power grid. On the other hand, EVs
42 (as ESSs) can be utilized to smoothen the power generation for the intermittent RESs and mitigate
43 their negative impacts on the power grid, such as voltage and frequency instability issues
44 (Karfopoulos & Hatziargyriou, 2016; Lam et al., 2016; A. A. S. Mohamed et al., 2016; Nguyen et
45 al., 2016; Rahbari-Asr et al., 2016; Sahoo et al., 2017; Wang et al., 2016). Among different RESs,
46 Photovoltaic (PV) generation shows the highest flexibility to be integrated with EV charging
47 stations due to the similarities between them (Elgammal & Sharaf, 2012; Hu & Liaw, 2016;
48 Kandasamy et al., 2017; Ahmed A. S. Mohamed, 2020; Ropp, 2009). Both PV and EVs will
49 probably be connected to the grid in the same highly dispersed manner and at the same voltage
50 levels. For example, EVs at residential homes are similar to those typical residential PV systems in
51 parameter values (Tomić & Kempton, 2007). Also, PV are more flexible to be installed at the
52 different EV charging locations, such as public charging stations, commercial buildings,
53 workplaces, and the vehicle itself (solar EV). In addition, both EV and PV are integrated to the grid
54 though power electronic interface with the potential to create intelligent nodes in the system. Due
55 to these common features between PV and EVs from the grid’s perspective, the grid-support
56 functions requested from PV will be similar to those required from EVs. In fact, it is possible that
57 codes and standards presently applied to the PV systems, such as IEEE-1547 and IEC-62116, may
58 apply to EV systems with minor modifications (Ropp, 2009). Considering the most commonly
59 used chargers [level 2 (L2) ac chargers], it is expected that both PV and L2 charging stations to be
60 integrated with the power grid through a common ac bus. In this case, extra challenges are added
61 to the grid-tie dc-ac inverter of the PV system. The inverter must be properly controlled to regulate
62 the dc-bus voltage, achieve synchronization with the grid, and minimize the power quality issues

2|Page
63 under different disturbances from both the generation (fluctuations of environmental conditions)
64 and load sides (sudden variations of EV loads), (Kandpal et al., 2016), (Nwaigwe et al., 2019), and
65 (Manias, 2017). Voltage source inverter (VSI) is typically used in grid-connected PV systems, and
66 controlled to manage the injected power to the grid with minimum impact on the grid stability and
67 power quality (Ahmed A. S. Mohamed et al., 2019), (Sakthi & Kowsalya, 2017), (Kharjule, 2015)
68 and (Ashourpouri et al., 2012). The VSI control system consists of an internal current loop, for grid
69 active and reactive current control, and an external voltage loop, for dc-link voltage regulation (Al-
70 Shetwi et al., 2018), (Vennelaganti et al., 2016) and (Agirman & Blasko, 2003). These control loops
71 are used to improve the system stability with slow dynamics so the dc-link voltage loop can be
72 used with an inner power loop so the current is not controlled directly (Kadri et al., 2011). Several
73 approaches for active and reactive power control are presented in the literature, such as sliding
74 mode (Chen et al., 2015), fuzzy logic control (Thao & Uchida, 2013), and predictive control method
75 (Golzari et al., 2019). However, these methods show sensitive performance to the variation in
76 power line inductance and switching frequencies. In (Heydari & Varjani, 2018), a direct power
77 control method was introduced that uses a fixed switching frequency. In (Gopalakrishnan &
78 Narayanan, 2014), the direct power with a space vector modulation has been applied for reducing
79 harmonics. The direct power control uses the instantaneous values of active and reactive power to
80 generate switching activities, which is challenges, especially in the existence of harmonics.

81 The most commonly used technique in industrial control is the classical proportional,
82 integral and derivative (PID) and PI control, due to the simplicity, and ease of implementation
83 (Lakshmi Sangeetha et al., 2012). In addition, classical controllers don’t require previous
84 knowledge for a plant characteristics (Yan et al., 2009). However, they change the control output
85 to minimize the error between an actual and desired reference value (Louzazni & Aroudam, 2014).
86 The main problem of PI controllers is the difficulty of finding the proper parameters that keep the
87 system operating at its optimum point. In addition, they sometimes require developing accurate
88 mathematical models for control design (Atherton, 1996). Several optimization techniques have
89 been presented in the literature for estimating and optimizing the PI controller parameters
90 (Annamraju & Nandiraju, 2018) and (Bernard & Musilek, 2017). Particle swarm optimization
91 (PSO) was introduced by (Gupta & Mahanty, 2015) to find the optimal values of control
92 parameters. In (Ebrahim et al., 2018), Genetic algorithm (GA) was considered for finding the PI
93 control parameters for an ac/dc microgrid. Salp swarm algorithm (SSA) shows successful
94 implementation in several applications, such as parameter optimization and target localization

3|Page
 95 (Kumari & Shankar, 2018), (Ekinci & Hekimoglu, 2018) and (Liu & Xu, 2018). It is easy to be
96 implemented, robust, and has a globe convergence capability (Mohapatra & Sahu, 2018). To the
97 authors’ knowledge, SSA has not been yet tested for PV grid-connected applications.
98 This paper presents a new tuning technique for the PI controller of the grid-tie dc-ac inverter
99 in grid-connected PV systems, supporting an EV charging station with ac L2 ports. The proposed
100 technique is based on analytical formulas combined with SSA optimization. The proposed
101 algorithm is used to find six gains associated with the dc-bus voltage regulator and current
102 controllers. The controller performance is validated under different dynamics associated with
103 uncertainties of EV charging loads and environmental conditions using simulation and
104 processor-in-the loop data. The performance of SSA-based controller is compared with another
105 one using particle swarm optimization (PSO) algorithm.
106 The paper is organized as follows: the mathematical modeling of the system under
107 consideration is presented in section 2; the utilization of the system model to analytically tune the
108 controller is discussed in section 3; in section 4, the SSA method is presented for optimizing the
109 controller’s parameters; section 5 presents estimation and prediction analysis for L2 ac charging
110 profile; the controller testing and validation are discussed in section 5; and finally the study is
111 concluded in section 6.

112 2. System Description and Modeling

113 A description for the system under consideration is depicted in Figure 1. It shows an ac
114 microgrid operates in a grid-connected mode. It contains a grid-connected PV power system
115 supplying a L2 EV charging station through a common ac bus. The PV system is coupled to the ac
116 bus through a dc-dc boost converter with maximum power point tracking (MPPT) capability, and
117 a three-phase VSI for managing the power-flow and regulating the dc-bus voltage for active
118 power-flow balance. The system under consideration experience several control challenges due to
119 the high level of uncertainties in the system from both the generation and load sides. The sudden
120 variations of radiation in cloudy days combined with the random and unexpected nature of the EV
121 charging process bring difficulties to the system operation. These fluctuations may cause
122 instability and power quality issues if they do not carefully be suppressed carefully by the power-
123 flow controller.

4|Page
 Utility grid AC bus

DC- DC PV array
Grid-tie
inverter boost converter

AC DC
L2 EV charging station
DC DC

Power MPPT
control

124
125 Figure 1: Block diagram for the grid-connected ac micro-grid with PV generator and EV charging station.

126 The configuration of the three-phase grid-tie VSI is indicated in Figure 2. The inverter output is
127 connected to an LC filter for managing the harmonics level. It is controlled to manage the active
128 and reactive power-flow and regulate the dc-link voltage. A nonlinear small signal model is
129 developed for the VSI to model the nonlinearities due to switching activities. Then, the model is
130 linearized to design a linear feedback control system. An average switched model is considered to
131 extract transfer functions needed for designing the control parameters. Park’s transformation is
132 applied to transform the system variables from the three-phase abc frame to the synchronous
133 rotating reference frame (dq). The dc quantities in the synchronous rotating reference frame are
134 used in a large signal model. Therefore, the modeling of VSI involves two power stage models:
135 large signal and small signal model.

136
137 Figure 2: Schematic diagram of three-phase VSI circuit with LC filter.

138 2.1. Power Stage Model

139 In power stage, the power circuit of the VSI inverter, in Figure 2, is transformed to the
140 synchronous rotating reference frame, where vinv_a, vinv_b, and vinv_c are the inverter output
141 voltages and vg_a ,vg_b, and vg_c are the grid voltages. Voltage balance equation is applied to the

5|Page
142 inverter output as follows:

_ _ _
143 _ = _ + _ , (1)
_ _ _

144 where L is the filter inductance with neglecting the internal resistance. Equation (1) is transformed
145 to the synchronous reference frame, as in equation (2) (Quintero et al., 2018).

−
= + +
_
_
146 (2)

147 Inverter voltages (vd_inv and vq_inv) and currents (id and iq) in the synchronous reference frame
148 are expressed as function of the dc-link voltage (vdc) and current (idc), as follows,

= = .
_
_
149 ; , (3)

150 where dd and dq are the dq components of the inverter duty cycle. Equations (2) and (3) are used
151 to find the power stage model as follows:

. = − + , " = +
! !_# $#
%
152 (4)

. = − + , " = −
$ $_# !#
%
153 (5)

154 & = ' () *+ + (1⁄ " *+ (6*

155 From equations (4)-(6), power stage model equivalent circuits are created as depicted in Figure 3.

156
157 Figure 3: Three phase VSI Power stage model (Messo et al., 2015).

158 2.2. Small Signal Model

6|Page
159 The small signal model is derived from the power stage model by assuming a small and linear
160 perturbation in the variables, considering a stable and fast system response. Equations (1)-(5) are
161 updated for small linear perturbations with neglecting the steady and non-linear terms as
162 follows:

= 0 −1 0 − 2 −
/0
!
163 (7)
0
" = 34 +
!# $#
%
164 (8)
165 3̂ = 1 34 + 1 34 + 2 +2 . (9)

166 Equations (7)-(9) are used for deriving the small signal model equivalent circuits shown in Figure 4.

167
168 Figure 4: Three phase VSI small signal model.

169 3. VSI Control Design Using Analytical Technique

170 The control of VSI involves three PI regulators for managing the d-axis current (id), q-axis
171 current (iq) and dc-bus voltage (vdc), as indicated in Figure 5. The dc-bus voltage regulator
172 generates a reference signal to the d-axis current (id*), which is compared to the actual measured
173 value (id) to generate the error signal (Δid), which is the input to the d-axis current PI regulator.
174 The reference signal for the q-axis current (iq*) is received externally for reactive power control
175 purposes. In this case, only active power control is achieved, so iq* is set to zero.

7|Page
 vdc Gf,dc
vdc* Δvdc dc-bus id *
voltage PI
regulator Gf,d
id Δid D-axis
current PI Demand
Measured abc control voltage
PWM
currents abc dq dq abc modulator
Δiq Q-axis
current PI
iq control
iq* Gf,q
176
177 Figure 5: VSI control in synchronous reference frame.

178 The small signal model displayed in Figure 4 is used to formulate the transfer functions that
179 control the d-axis current (Gf,d), q-axis current (Gf,q), and dc voltage (Gf,dc) (Puukko et al., 2012).
180 To simplify the derivation, the grid voltage is oriented along the d-axis, such that vdg is constant
181 and vqg equals to zero. The open loop transfer function in S-domain are derived based on the small
182 signal models in the frequency domain, as follows:

1 6 + 7 = 3̂ + 83̂ + 6 9 = =
/̂! !;
:! <=
183 (10)

+ 7
/̂$
1 6 = 3̂ + 83̂ + 6 9 = :$
= !;
<=
184 (11)

3̂ = 1 3̂ + 1 3̂ + 7 > + 7 > 9 = = +
6 !; ?! A! <
/̂! =@!; @!; !;
185 . (12)

186 The final transfer functions for the three PI regulators of the VSI in terms of the open loop
187 transfer functions are presented in equations (13)-(15). These functions are used to find the
188 optimum gains for the PI regulators.
EF,! GEH,! ×I
189 9B, = 9 × I
(13)
EF,$ GEH,$ ×I
9B, = 9 ×
I
190 (14)
EF,!; GEH,!; ×I
9B, =9 ×
I
191 (15)

192 In general, the behavior of a closed-loop control system is approximated by the second order
193 reference model presented in equation (16).

JK (L* = =O G+×P×MN
MO
Q ×IGMN
194 O (16)

195 where, ω0 determines the angular speed response, and ζ denotes the damping coefficient. The
196 open loop transfer function H(s) is obtained as follows:

8|Page
 R (I*
J(L* = STR =
N MNO
N (I* =(=G+×P×MQ *
197 . (17)

198 Applying the model in equation (17) to the VSI transfer functions in equations (13)-(15), provides:

MO
9B, = =(=G+×PN!×M
N! *
(18)
!
199

MN$
O
9B, =
=(=G+×P$ ×MN$ *
200 (19)
O
9B, =
MN!;
=(=G+×P!; ×MN!; *
201 (20)

202 where, wod, Ud, woq, Uq, wodc, and Udc are the speed response and damping value of d-axis current,
203 q-axis current and dc bus voltage control systems.

204 Using equations (13-15) and (18-20), the gains kp,d, ki,d, kp,q, ki,q, kp,dc, and ki,dc are extracted to
205 provide a predefined performance conditions (wod, Ud, woq, Uq, wodc, and Udc). These conditions are
206 related to the pole-zero patterns of each PI regulator, which are implemented in the sisotool
207 Matlab command in the control system toolbox. The tool provides a graphical user interface (GUI)
208 where the closed-loop frequency response can be changed by modifying the pole-zero pattern of
209 the feedback controller, as shown in Figures 6 and 7. sisotool function is used with equations (10)-
210 (12) for obtaining the PI parameters that show minimum rise time with zero overshoot. The dq
211 axis current controller gains are obtained by online adjustment for poles-zeroes, as indicated in
212 Figure 6. While the parameter tuning of the dc voltage regulator is depicted in Figure 7.
213

214
215 Figure 6: PI regulator performance for d and q axes currents: (a) root locus editor, (b) Nichols editors, (c) bode editor,
216 and (d) step response.
217
218

9|Page
219
220 Figure 7: PI regulator performance for DC voltage: (a) root locus editor, (b) Nichols editors, (c) bode editor, and (d)
221 step response.
222 The final optimum gains are presented in Table 1. These gains show a 11.5 µsec rise time, a 19.5
223 µsec settling time and a 0.02 v steady state error (see Figures 6 and 7). PI gains are equal for d-
224 axis and q-axis currents because of the equality of Gd and Gq.

225 Table 1: PI compensator gains based on analytical small signal model

226 Gain Value
227 kp,d and kp,q 0.03
ki,d and ki,q 7
228 kp,dc 2
229 ki,dc 100
230
231
232 4. Control Design Using SSA Optimization.

233 SSA optimization shows several advantages over the other techniques due to the ease of
234 implementation, robustness, and globe convergence capability (Mirjalili et al., 2017). SSA is
235 considered in this work for tuning kps and kis parameters by searching for the optimum values
236 that show minimum steady-state errors of d-axis current, q-axis current, and dc-bus voltage
237 signals. The mean square errors (MSE) is considered as the objective function, and the six gains
238 (kp,d, ki,d, kp,q, ki,q, kp,dc, and ki,dc) are tuned to minimize this objective, as indicated in Figure 8.

10 | P a g e
 Proposed tuning approach based on SSA

kp,dc ki,dc k p,d k i,d k p,q k i,q

Current controller vdg

Voltage regulator
* i*
kp,dc ki,dc kp,d ki,d id kp,q ki,q iq
vdc d vqg
* vdg vqg
vdc id PI id PI
PI
vdc
i*
q = 0 Pulses PWM modulator

id iq

PWM
iPV dq PLL
dq
MPPT abc abc
VPV
VSI Grid
DC-DC La ~

vg_a
i inv_a
boost
converter
C
dc Lb ~

vg_b
i inv_b
PV
Lc ~

vg_c
i inv_c
Voltage signal Ra Rb Rc
Current signal
Ca Cb Cc EV station
Phase shift value
239 Voltage sensor Filter
240 Figure 8: PV grid-connected system with the proposed control of VSI.

241 SSA is inspired from the salps jelly fishes, which their move are affected by pumped water
242 through their transparent body (Anderson et al., 1980). Their swarming behavior in deep oceans
243 is formed a salp chain form to coordinate their motion rapidly with pumped water. Salps chain is
244 expressed mathematically by two groups: leaders and followers. Leaders present in the front of a
245 chain to guide the salps, while followers follow each other (Andersen & Nival, 1986).
246 A salp position is represented by n-dimensional search space, where n is the number of search
247 variables. All positions are represented in a two-dimensional matrix called x. The food source (F)
248 is used in the search space as the swarm’s target. For updating the position of the leader with
249 respect to the food source, the following equation is considered (Mirjalili et al., 2017):

YW + "S Z"+ [\]W − ^]W _ + ^]W ` "a ≥ 0

VWS = X
YW − "S Z"+ [\]W − ^]W _ + ^]W ` "a < 0
250 (21)

251 where VWS represents the position of the first salp (leader) in the j-th dimension; Fj is the position
252 of the food source in the j-th dimension, \]W and ^]W indicate the upper and lower bound of j-th
253 dimension, respectively; and C1, C2, and C3 are the algorithm coefficients. The C1 is the most
254 important parameter, which balances the exploration and exploitation process and estimated as

11 | P a g e
255 follows (Mirjalili et al., 2017):
gh O
256 "S = 2f T( i * (22)

257 where m is the current iteration and M is the maximum number of iterations. The coefficients C2
258 and C3 are random numbers, which are uniformly generated within the interval [0, 1] to predict
259 the next position. The next position is determined in positive or negative infinity with j-th
260 dimension as well as step size based on these parameters. The followers’ positions are updated as
261 follows based on Newton’s law of motion:

VW = + jk + + l k; where Є 2, s
S
262 (23)

263 where VW shows the position of i-th follower salp in j-th dimension, t is the time, v0 is the initial
264 speed, and a is the speed ratio, as indicated in equation (24).

j= ; where =
tFuvw yTyQ
B x (24)
Q
265

266 Because the time in optimization is represented by iteration, the discrepancy between
267 iterations is equal to 1 and, considering v0=0, so equation (23) can be expressed as follows:

VW = + (VW + VW TS * , where i ≥ 2
S
268 (25)

269 Using equations (21) and (25), the salp chains can be tested, based on the algorithm presented
270 in Figure 9.

12 | P a g e
271
272 Figure 9: SSA optimization algorithm.

273

274 SSA is implemented and used for finding the best combination of PI gains. The MSE for the PI
275 regulators output signals [Equation (26)] is selected as objective function to observe the error
276 performance for improving both transient and steady-state responses of the inverter current.

z8{ = ∑ ~S(∆ +∆ +∆ *
S + + +
277 (26)

278 Where i is the current sample, and n is the maximum number of samples. The settings of SSA
279 are presented in Table 2. Six gains for the three PI regulators are optimized, considering a
280 maximum number of iterations of 100, with 10 samples each based on the search agents, so the
281 total number of iterations is 1000. The lower boundaries are set to zero to open the search space
282 for the optimizer, and the settings of the upper boundaries are set by trial and error.

283 TABLE 2: SSA settings

Parameter Value
Number of variables 6
Variables [kp,d, ki,d, kp,q, ki,q, kp,dc, ki,dc]
Number of search agents 10

13 | P a g e
 Maximum number of iterations 100
Lower bound [0, 0, 0, 0, 0, 0]
Upper bound [1, 20, 1, 20, 10, 1000]

284 SSA-based optimizer is applied to the system model in Figure 8, considering the parameters
285 in Table 2. The optimization performance is indicated in Figure 10, which shows the progression
286 of the best fitness value. As it can be noticed, a significant reduction in the fitness value is
287 experienced at the beginning and a best value of 1.05x10-4 is settled around iteration 800. The
288 behavior of the search variables (PI gains) during the optimization process is shown in Figure 11.
289 The gains realize very small variation after iteration 800, due to realizing the best fitness. The
290 final optimum gains are presented in Table 3.

291
292 Figure 10: Best score of SSA for fitness function

293
294 Figure 11: SSA tuning process (kp,d, ki,d, kp,q, ki,q, kp,dc, ki,dc)
295 Table 3: SSA based optimal PI gains
Gain Value
kp,d 0.25088
ki,d 14.9946
kp,q 0.44803

14 | P a g e
296 ki,q 12.1020
kp,dc 4.7349
297
ki,dc 461.37
298

299

300

301 5. Aggregated Load Profile of EV Charging Station

302 In this section, the power profile for EV charging station is generated based on assumptions
303 and probability analysis. The station is assumed to be located at a parking garage at the
304 engineering campus at Zagazig University, Zagazig, Egypt. It contains 20 level 2 ac charging
305 stations with 6.6 kW maximum charging rate capabilities. The charging load prediction is based
306 on a probabilistic approach for the number of cars charged with time. A probability density
307 function with normal distribution during the day is assumed, as presented in equation (27).
†
T
ˆ‰Š O
•(€; μ, ‚ + * = ƒ√+… f
S O‡ Œ
308 ‹ (27)

309 Where f is probability density function of the number of EVs, y is the time step, µ and σ are the
310 mean and the standard deviation.

311 Typically, in the faculty of engineering, most of the classes run on weekdays in the morning,
312 which leads to a peak demand around 10 A.M., as indicated in Figure 12. In this case, around 15
313 cars may charge at the same time. The campus may realize different behavior during exams,
314 which has two sessions one in the morning and another one in the afternoon. In these days, two
315 peaks may occur at charging station during the day with an expected maximum peak at noon,
316 during the transition between the two sessions, as depicted in Figure 13.

317
318 Figure 12: Typical daytime (8 A.M. to 5 P.M.) profile. (a) Probability density. (b) Number of charging EVs.

15 | P a g e
319
320 Figure 13: Special daytime (8 A.M. to 5 P.M.) profile. (a) Probability density. (b) Number of charging EVs.

321 The data is discretized, and a factor of variation is introduced Figures 12(a) and 13(a). The
322 weight, mean, and standard deviation parameters of the function are obtained based on the usual
323 class schedule, the day of the week, and the time of the day. The hypothesis adopted is that EVs
324 only operate in charging mode with unregulated 6.6 kW charging rate.
325

326 6. Results and Discussion.

327 A 120 kW grid-connected PV system is constructed in Matlab/Simulink based on the

328 equivalent circuits, shown in Figures 2 and 8. It consists of a PV array with a single diode model, a
329 dc-dc boost converter with MPPT, and a three-phase VSI, which is interfaced with the utility
330 through an LC filter. In addition, the EV charging profile, described in section 5, is connected to
331 the common ac bus. The system parameters are presented in Table 4. The VSI is driven by the
332 proposed controller and its performance is evaluated in this section under different environmental
333 and loading conditions.

334 Table 4: Parameters of the PV system components

Parameter Value
PV module model First Solar FS-272
Rs, Rsh, and Iph 7.77 Ω, 1044.9 Ω, and 1.19 A
Cpv (µF) 100
L1 (mH) 5
Cdc (µF) 12,000
Lfilter (µH) 250
Cfilter (µF) 125
Rfilter (mΩ) 2

335 Two real-world irradiance profiles representing two days in Egypt: sunny and cloudy, shown in
336 Figure 14, are investigated. These data are collected by the Electronics Research Institute (ERI),
337 Giza, Egypt at February 2015 (Elkholy et al., 2016). In addition, two EV charging profiles, shown

16 | P a g e
338 in Figure 15, are considered. These profiles represent the two cases described in section 5,
339 considering aggregated EVs forecasting in a typical weekday (one-peak profile) and a special
340 weekday (two-peaks profile). The proposed optimal controller based on SSA is assessed under all
341 these dynamics and compared with the baseline analytical design. For analytical design, the gains
342 in Table 1 are considered, while for the optimized design based on SSA, the gains in Table 3 are
343 used.
(b) (a)
400
800
600
200 400
200
0 0
6:30 8:00 10:00 12:00 14:00 16:00 18:30 6:30 8:00 10:00 12:00 14:00 16:00 18:30
Time [h:mm] Time [h:mm]
344
345 Figure 14: Irradiance profiles. (a) Cloudy day. (b) Sunny day.

346
347 Figure 15: EV charging load profiles. (a) One peak profile. (b) Two peaks profile.

348 a. Performance of the SSA- and Analytical-based Controllers

349 The impact of the proposed controllers on the dc-bus voltage waveform under different
350 irradiance and load profiles is indicated in Figure 16. Both analytical- and SSA-based controllers
351 are tested considering four cases of dynamics: case 1 represents a sunny-day radiation and one-
352 peak EV loading profile, case 2 shows a sunny-day radiation and two peaks EV loading profile,
353 case 3 tests a cloudy-day radiation and one peak EV loading profile, and finally case 4 investigates
354 cloudy radiation and two peaks EV loading profiles. The reference dc-bus voltage is chosen as 500
355 V to verify the peak output value of ac inverter voltage. As it can be noticed, in all cases, the SSA-
356 based controller shows better tracking performance for the reference value compared to the
357 analytical-based one. The largest disturbance in the dc-bus voltage is noticed in cases 2 and 4
358 [Figure 16(b) and 16 (d)], which deploy two-peaks load profile. The second peak has the most

17 | P a g e
359 negative impact due to its impulse nature. The SSA-based controller shows very fast suppression
360 to the variation of the dc-bus voltage during this peak compared to the analytical-based controller.
361

362
363 Figure 16: dc-bus voltage profiles under SSA- and analytical-based controllers. (a) Case 1. (b) Case 2. (c) Case 3. (d)
364 Case 4.

365 Other indices that measure the performance of the controller are the error signals of d-axis
366 current, q-axis current, and dc-bus voltage, which are the main objectives of the SSA optimization.
367 These error signals are displayed in Figures 17 and 18, for analytical- and SSA-based controllers,
368 respectively. Each controller is evaluated considering the four cases of generation and loading
369 profiles mentioned earlier. Both controllers try to reduce the errors to zero. For the worst case
370 (case 4), the SSA-based controller shows MSE of 0.0128 for the d-axis current, 0.00145 for the q-
371 axis current, 0.0017 for the dc-bus voltage. In addition, the peak to peak ripple of the dc-bus
372 voltage (vdc_ripple) shows 10% of the reference value. On the other hand, the analytical-based
373 controller realizes MSE of 0.0319 for the d-axis current, 0.0064 for the q-axis current, and 0.0035
374 for the dc-bus voltage, and 19.3% peak to peak ripple of the dc-bus voltage. The MSE for all signals
375 under different cases of operating conditions along with vdc_ripple are presented numerically in Table
376 5. In addition, other performance indices for d-axis current, q-axis current, and dc bus voltage
377 signals, such as settling time (Ts) in millisecond, percentage overshoot (PO), and percentage

18 | P a g e
378 undershoot (PU) are estimated for the two controllers at the four operation cases, as presented in
379 Table 6.

380
381 Figure 17: Error signals with analytical-based controller. (a) Case 1. (b) Case 2. (c) Case 3. (d) Case 4.

382
383 Figure 18: Error signals with SSA-based controller. (a) Case 1. (b) Case 2. (c) Case 3. (d) Case 4.

384 Table 5: MSE and vdc_ripple signals with Anaytical- and SSA-based controllers
Case Controller ∆id ∆iq ∆vdc vdc_ripple (%)
Analytical 0.0238 0.0071 0.0026 11.7
Case 1
SSA 0.0135 0.001 0.0014 4.3
Analytical 0.04134 0.0082 0.00397 17
Case 2
SSA 0.0153 0.0023 0.00191 9.2
Analytical 0.0245 0.00695 0.0027 12.2
Case 3
SSA 0.0122 0.0012 0.0016 4.7

19 | P a g e
 Analytical 0.0319 0.0064 0.0035 19.3
Case 4
SSA 0.0128 0.00145 0.0017 10

385 Table 6: Settling time, overshoot, and undershoot with Analytical- and SSA-based controllers
Case Parameter Controller Ts (msec) PO (%) PU (%)
Analytical 78 0.13 0.04825
id
SSA 28.7 0.1075 0.0325
Analytical 50 0.05125 0.04375
Case1 iq
SSA 7.1 0.0465 0.035
Analytical 112 6.2 5.54
vdc
SSA 11.7 2.1 2.2
Analytical 332.71 0.29125 0.0905
id
SSA 21.2 0.25475 0.03975
Analytical 236.8 0.126 0.064
Case2 iq
SSA 14.4 0.1145 0.05
Analytical 375 3 14
vdc
SSA 29.63 1.84 7.32
Analytical 50 0.12575 0.03125
id
SSA 13.4 0.1115 0.0275
Analytical 9.4 0.05125 0.024
Case3 iq
SSA 6.98 0.05 0.0225
Analytical 99.5 5.8 6.2
vdc
SSA 20.2 2.3 2.44
Analytical 201.2 0.30125 0.0525
id
SSA 20.1 0.275 0.0225
Analytical 20 0.1275 0.0675
Case4 iq
SSA 14 0.1125 0.065
Analytical 201.2 4.3 15
vdc
SSA 22.1 2 8
386

387 Considering the worst-case operating scenario with cloudy-day and two-peaks load profiles,
388 the inverter output voltage and current waveforms and fast-Fourier transform (FFT) distribution

20 | P a g e
389 are investigated in Figures 19 and 20, respectively. The analytical-based controller shows a total
390 harmonic distortion (THD) for the voltage of 0.82% and for the current of 4.5%. The SSA-based
391 controller reduces the THD of voltage to 0.49% and of current to 1.63%.

392
393 Figure 19: Inverter output voltage waveforms and FFT. (a) Analytical. (b) SSA.

21 | P a g e
394
395 Figure 20: Inverter output current waveforms and FFT. (a) Analytical. (b) SSA.

396 The power profiles of inverter output (Pinv), grid (Pgrid), and EV loads (Pload) under different
397 operating scenarios are indicated in Figure 21. For sunny days (cases 1 and 2), the PV generation is
398 enough to supply the majority of the charging loads most of the day. Only minor support from the
399 grid is needed during the second peak of charging profile (case 2). In cloudy days, most of the
400 charging loads are supplied by the grid (cases 3 and 4).

401
402 Figure 21: Power-flow profiles for inverter, load and grid considering SSA-based controller. (a) Case 1. (b) Case 2. (c)
403 Case 3. (d) Case 4.

22 | P a g e
404 b. Performance of the SSA- and PSO-based Controllers

405 The SSA optimization algorithm shows similarities with PSO algorithm, which encourages
406 researchers to compare between both algorithms in several applications, such as medical,
407 computer science, and economics (Khan et al., 2019), (Syed & Syed, 2019) and (Hegazy et al.,
408 2020). Similarly, in this work, the proposed controller is optimized by PSO algorithm and
409 compared with the SSA-based controller. The conventional PSO algorithm presented in (Eberhart
410 & Kennedy, 1995) is considered in this study. The settings for the PSO solver are presented in
411 Table 7. The Six gains of PI regulators are optimized, considering a maximum number of iterations
412 of 100, with 10 samples per iteration based on the swarm size, so the total number of iterations is
413 1000. The lower and upper boundaries for the search variables are same as those of SSA solver.
414 Considering PSO solver, the final optimum gains for the PI regulators are: kpd=0.49642,
415 kid=7.7834, kpq=0.37473, kiq=14.5933, kpdc=3.9972, and kidc=364.3148. The final best fitness value
416 is 0.0041134, which is higher than that of SSA (1.05x10-4). The PSO-based controller is tested for
417 the worst case (case 4), and compared with the SSA-based one, as indicated in Figure 22, which
418 shows the dc-bus voltage. As it can be noticed, the SSA-based controller shows less variation in the
419 dc-bus voltage compared to the PSO-based one during both transient and steady-state operation.
420
421 TABLE 7: PSO settings
Parameter Value
Number of variables 6
Variables [kp,d, ki,d, kp,q, ki,q, kp,dc, ki,dc]
Swarm size 10
Maximum number of iterations 100
Lower bound [0, 0, 0, 0, 0, 0]
Upper bound [1, 20, 1, 20, 10, 1000]

422
423 Figure 22: dc-bus voltage profiles for SSA- and PSO-based controllers during Case 4 operation.

424

23 | P a g e
425 For case 4 operating scenario, the PSO-based controller shows MSE of 0.023 for the d-axis
426 current, 0.00159 for the q-axis current, and 0.0021 for the dc-bus voltage, and dc-bus voltage
427 ripple of 11.9%. These values are higher than those for the SSA-based controller presented in Table
428 5. Other performance indices are evaluated for both PSO- and SSA-based controllers during case 4
429 and presented in Table 8. As it can be seen, the SSA-based controller shows better control
430 performance indicated by the lower settling time, PO and PU.

431 Table 8: Settling time, overshoot, and undershoot with PSO- and SSA-based controllers
Parameter Controller Ts (msec) PO (%) PU (%)
PSO 32 0.28025 0.04325
id
SSA 20.1 0.275 0.0225
PSO 15.5 0.12075 0.065
iq
SSA 14 0.1125 0.065
PSO 35.8 2.3 9.6
vdc
SSA 22.1 2 8

432 c. Processor-in-the-Loop (PIL) Implementation

433 The proposed SSA-based controller is validated using the processor-in-the-loop (PIL) concept.
434 The PI regulators are embedded on a microcontroller board Arduino UNO REV3, while the PWM
435 generation along with the power components are represented in MATLAB/Simulink environment,
436 as indicated in Figure 23. The microcontroller board is programmed using Arduino IDE 1.8.8
437 software. The Arduino UNO REV3 receives analog signals for the dc-bus voltage reference (vdc*)
438 and the q-axis current reference (iq*) together with the actual measured values vdc, iq, and id. It
439 generates vd_ref and vq_ref, and sends them as analog signals to the PWM modulator.
440

24 | P a g e
441
442 Figure 23: Processor-in-the-loop (PIL) implementation using Arduino UNO REV3 board.

443 The PIL platform in Figure 23 is used for testing the proposed SSA-based controller during
444 case 4 operation, as a worst-case scenario. The results from the PIL platform are compared with
445 the simulation results, as depicted in Figure 24, which presents the output signal of the three PI
446 regulators. As it can be noticed, the PIL test shows a little higher error than the simulation data,
447 with MSE of 0.02 for d-axis current, 0.0025 for q-axis current, 0.0025 for dc-bus voltage, THD for
448 the voltage of 0.65%, and for the current of 2%. In addition, the real-time PIL test realizes slower
449 performance due to the delay in the communication signals. However, in general, good correlation
450 can be observed between both simulated and experimental results, which ensure the ability of
451 implementing the proposed controller in real-time.

25 | P a g e
452
453 Figure 24: Comparison between PIL and simulation results for case 4 operation. (a) d-axis current PI regulator output.
454 (b) q-axis current PI regulator output. (c) dc bus voltage PI regulator output.

455 7. Conclusion

456 In this paper presented control design optimization methodologies for a three-phase VSI
457 connecting a PV power system with an ac EV charging station. The controller’s parameters are
458 optimized using three techniques: analytical, SSA, and PSO. The optimization objectives include
459 minimum fluctuation in the dc-bus voltage and reference d-axis and q-axis currents, represented
460 by the mean-square error in the signals. A hypothetical EV charging station with 20 ac L2 chargers
461 is modeled under different operating conditions and considered for testing the controller
462 robustness. In addition, real-world irradiance profiles representing sunny and cloudy days are
463 integrated with the PV system model. The proposed SSA-based controller is implemented in real-
464 time using PIL concept. The results show that the SSA-based controller was able to minimize the
465 error in dc-bus voltage by about 50% compared the analytical-based one. Furthermore, the impact
466 of the controller on the injected harmonics is explored, which shows that the SSA-based design
467 provides less voltage and current harmonics and THD compared to the analytical one.

26 | P a g e
468 Acknowledgment

469 The National Renewable Energy Laboratory (NREL) is the current address for the first author
470 only. NREL and the U.S. Department of Energy (DOE) did not contribute to this work.

471 References

472 Agirman, I., & Blasko, V. (2003). A novel control method of a VSC without AC line voltage sensors.
473 IEEE Transactions on Industry Applications, 39(2), 519–524.
474 https://doi.org/10.1109/TIA.2003.808925
475 Al-Shetwi, A. Q., Sujod, M. Z., & Blaabjerg, F. (2018). Low voltage ride-through capability control
476 for single-stage inverter-based grid-connected photovoltaic power plant. Solar Energy, 159,
477 665–681. https://doi.org/10.1016/j.solener.2017.11.027
478 Andersen, V., & Nival, P. (1986). A model of the population dynamics of salps in coastal waters of
479 the Ligurian Sea. Journal of Plankton Research, 8(6), 1091–1110.
480 https://doi.org/10.1093/plankt/8.6.1091
481 Anderson, P. a. V., Bone, Q., & Denton, E. J. (1980). Communication between individuals in salp
482 chains. II. Physiology. Proceedings of the Royal Society of London. Series B. Biological
483 Sciences, 210(1181), 559–574. https://doi.org/10.1098/rspb.1980.0153
484 Annamraju, A., & Nandiraju, S. (2018). Frequency Control in an Autonomous Two-area Hybrid
485 Microgrid using Grasshopper Optimization based Robust PID Controller. 2018 8th IEEE
486 India International Conference on Power Electronics (IICPE), 1–6.
487 https://doi.org/10.1109/IICPE.2018.8709428
488 Ashourpouri, A., Sheikholeslami, A., Shahabi, M., & Niaki, S. A. N. (2012). Active power control of
489 smart grids Using Plug-in Hybrid Electric Vehicle. Iranian Conference on Smart Grids, 1–
490 6.
491 Atherton, D. P. (1996). Some advantages of the autotuning approach in PID control. IEE
492 Colloquium on Getting the Best Our of PID in Machine Control, 1/1-1/6.
493 https://doi.org/10.1049/ic:19961460
494 Bernard, M., & Musilek, P. (2017). Ant-based optimal tuning of PID controllers for load frequency
495 control in power systems. 2017 IEEE Electrical Power and Energy Conference (EPEC), 1–
496 6. https://doi.org/10.1109/EPEC.2017.8286152
497 Chen, Z., Luo, A., Wang, H., Chen, Y., Li, M., & Huang, Y. (2015). Adaptive sliding-mode voltage
498 control for inverter operating in islanded mode in microgrid. International Journal of

27 | P a g e
499 Electrical Power & Energy Systems, 66, 133–143.
500 https://doi.org/10.1016/j.ijepes.2014.10.054
501 Eberhart, R., & Kennedy, J. (1995). A new optimizer using particle swarm theory. MHS’95.
502 Proceedings of the Sixth International Symposium on Micro Machine and Human Science,
503 39–43. https://doi.org/10.1109/MHS.1995.494215
504 Ebrahim, Ahmed. F., Mohamed, A. A. S., Saad, A. A., & Mohammed, O. A. (2018). Vector
505 Decoupling Control Design Based on Genetic Algorithm for a Residential Microgrid System
506 for Future City Houses at Islanding Operation. SoutheastCon 2018, 1–5.
507 https://doi.org/10.1109/SECON.2018.8479013
508 Ekinci, S., & Hekimoglu, B. (2018). Parameter optimization of power system stabilizer via Salp
509 Swarm algorithm. 2018 5th International Conference on Electrical and Electronic
510 Engineering (ICEEE), 143–147. https://doi.org/10.1109/ICEEE2.2018.8391318
511 Elgammal, A. A. A., & Sharaf, A. M. (2012). Self-regulating particle swarm optimised controller for
512 (photovoltaic-fuel cell) battery charging of hybrid electric vehicles. IET Electrical Systems
513 in Transportation, 2(2), 77–89. https://doi.org/10.1049/iet-est.2011.0021
514 Elkholy, A., Fahmy, F. H., Abou El-Ela, A. A., Nafeh, A. E.-S. A., & Spea, S. R. (2016).
515 Experimental evaluation of 8kW grid-connected photovoltaic system in Egypt. Journal of
516 Electrical Systems and Information Technology, 3(2), 217–229.
517 https://doi.org/10.1016/j.jesit.2015.10.004
518 Golzari, S., Rashidi, F., & Farahani, H. F. (2019). A Lyapunov function based model predictive
519 control for three phase grid connected photovoltaic converters. Solar Energy, 181, 222–
520 233. https://doi.org/10.1016/j.solener.2019.02.005
521 Gopalakrishnan, K. S., & Narayanan, G. (2014). Space vector based modulation scheme for
522 reducing capacitor RMS current in three-level diode-clamped inverter. Electric Power
523 Systems Research, 117, 1–13. https://doi.org/10.1016/j.epsr.2014.06.025
524 Gupta, V. K., & Mahanty, R. (2015). Optimized switching scheme of cascaded H-bridge multilevel
525 inverter using PSO. International Journal of Electrical Power & Energy Systems, 64, 699–
526 707. https://doi.org/10.1016/j.ijepes.2014.07.072
527 Hegazy, Ah. E., Makhlouf, M. A., & El-Tawel, Gh. S. (2020). Improved salp swarm algorithm for
528 feature selection. Journal of King Saud University - Computer and Information Sciences,
529 32(3), 335–344. https://doi.org/10.1016/j.jksuci.2018.06.003

28 | P a g e
530 Heydari, E., & Varjani, A. Y. (2018). Combined modified P amp;O algorithm with improved direct
531 power control method applied to single-stage three-phase grid-connected PV system. 2018
532 9th Annual Power Electronics, Drives Systems and Technologies Conference (PEDSTC),
533 347–351. https://doi.org/10.1109/PEDSTC.2018.8343821
534 Hu, K. W., & Liaw, C. M. (2016). Incorporated Operation Control of DC Microgrid and Electric
535 Vehicle. IEEE Transactions on Industrial Electronics, 63(1), 202–215.
536 https://doi.org/10.1109/TIE.2015.2480750
537 Kadri, R., Gaubert, J.-P., & Champenois, G. (2011). An Improved Maximum Power Point Tracking
538 for Photovoltaic Grid-Connected Inverter Based on Voltage-Oriented Control. IEEE
539 Transactions on Industrial Electronics, 58(1), 66–75.
540 https://doi.org/10.1109/TIE.2010.2044733
541 Kandasamy, N. K., Kandasamy, K., & Tseng, K. J. (2017). Loss-of-life investigation of EV batteries
542 used as smart energy storage for commercial building-based solar photovoltaic systems.
543 IET Electrical Systems in Transportation, 7(3), 223–229. https://doi.org/10.1049/iet-
544 est.2016.0056
545 Kandpal, M., Hussain, I., & Singh, B. (2016). Grid integration of single source SPV system using
546 asymmetric cascaded 7-level VSC. 2016 IEEE 6th International Conference on Power
547 Systems (ICPS), 1–6. https://doi.org/10.1109/ICPES.2016.7584012
548 Karfopoulos, E. L., & Hatziargyriou, N. D. (2016). Distributed Coordination of Electric Vehicles
549 Providing V2G Services. IEEE Transactions on Power Systems, 31(1), 329–338.
550 https://doi.org/10.1109/TPWRS.2015.2395723
551 Karfopoulos, E. L., Voumvoulakis, E. M., & Hatziargyriou, N. (2014). Steady-state and dynamic
552 impact analysis of the large scale integration of plug-in EV to the operation of the
553 autonomous power system of Crete Island. MedPower 2014, 1–9.
554 https://doi.org/10.1049/cp.2014.1648
555 Khan, I. A., Alghamdi, A. S., Jumani, T. A., Alamgir, A., Awan, A. B., & Khidrani, A. (2019). Salp
556 Swarm Optimization Algorithm-Based Fractional Order PID Controller for Dynamic
557 Response and Stability Enhancement of an Automatic Voltage Regulator System.
558 Electronics, 8(12), 1472. https://doi.org/10.3390/electronics8121472
559 Kharjule, S. (2015). Voltage source inverter. 2015 International Conference on Energy Systems
560 and Applications, 537–542. https://doi.org/10.1109/ICESA.2015.7503407

29 | P a g e
561 Kumari, S., & Shankar, G. (2018). A Novel Application of Salp Swarm Algorithm in Load
562 Frequency Control of Multi-Area Power System. 2018 IEEE International Conference on
563 Power Electronics, Drives and Energy Systems (PEDES), 1–5.
564 https://doi.org/10.1109/PEDES.2018.8707635
565 Lakshmi Sangeetha, A., Naveenkumar, B., Balaji Ganesh, A., & Bharathi, N. (2012). Experimental
566 validation of PID based cascade control system through SCADA–PLC–OPC and internet
567 architectures. Measurement, 45(4), 643–649.
568 https://doi.org/10.1016/j.measurement.2012.01.005
569 Lam, A. Y. S., Leung, K. C., & Li, V. O. K. (2016). Capacity Estimation for Vehicle-to-Grid
570 Frequency Regulation Services With Smart Charging Mechanism. IEEE Transactions on
571 Smart Grid, 7(1), 156–166. https://doi.org/10.1109/TSG.2015.2436901
572 Li, C., Cao, Y., Kuang, Y., & Zhou, B. (2016). Influences of Electric Vehicles on Power System and
573 Key Technologies of Vehicle-to-Grid. Springer.
574 Liu, X., & Xu, H. (2018). Application on Target Localization Based on Salp Swarm Algorithm. 2018
575 37th Chinese Control Conference (CCC), 4542–4545.
576 https://doi.org/10.23919/ChiCC.2018.8482543
577 Louzazni, M., & Aroudam, E. (2014). Control and stabilization of three-phase grid connected
578 photovoltaics using PID-Fuzzy logic. 2014 IEEE International Conference on Intelligent
579 Energy and Power Systems (IEPS), 279–284. https://doi.org/10.1109/IEPS.2014.6874195
580 Manias, S. N. (2017). 6—Inverters (DC–AC Converters). In S. N. Manias (Ed.), Power Electronics
581 and Motor Drive Systems (pp. 271–500). Academic Press. https://doi.org/10.1016/B978-
582 0-12-811798-9.00006-8
583 Messo, T., Aapro, A., & Suntio, T. (2015). Generalized multivariable small-signal model of three-
584 phase grid-connected inverter in DQ-domain. 2015 IEEE 16th Workshop on Control and
585 Modeling for Power Electronics (COMPEL), 1–8.
586 https://doi.org/10.1109/COMPEL.2015.7236460
587 Mirjalili, S., Gandomi, A. H., Mirjalili, S. Z., Saremi, S., Faris, H., & Mirjalili, S. M. (2017). Salp
588 Swarm Algorithm: A bio-inspired optimizer for engineering design problems. Advances in
589 Engineering Software, 114, 163–191. https://doi.org/10.1016/j.advengsoft.2017.07.002
590 Moeini-Aghtaie, M., Farzin, H., Fotuhi-Firuzabad, M., & Amrollahi, R. (2017). Generalized
591 Analytical Approach to Assess Reliability of Renewable-Based Energy Hubs. IEEE

30 | P a g e
592 Transactions on Power Systems, 32(1), 368–377.
593 https://doi.org/10.1109/TPWRS.2016.2549747
594 Mohamed, A. A. S., Berzoy, A., & Mohammed, O. (2016). Design and Hardware Implementation of
595 FL MPPT Control for PV Systems Based on GA and Small-Signal Analysis. IEEE
596 Transactions on Sustainable Energy, PP(99), 1–1.
597 https://doi.org/10.1109/TSTE.2016.2598240
598 Mohamed, Ahmed A. S. (2020). Dynamic Modeling Analysis of Direct-Coupled Photovoltaic
599 Power Systems. In A. M. Eltamaly & A. Y. Abdelaziz (Eds.), Modern Maximum Power Point
600 Tracking Techniques for Photovoltaic Energy Systems (pp. 439–461). Springer
601 International Publishing. https://doi.org/10.1007/978-3-030-05578-3_17
602 Mohamed, Ahmed A. S., Metwally, H., El-Sayed, A., & Selem, S. I. (2019). Predictive neural
603 network based adaptive controller for grid-connected PV systems supplying pulse-load.
604 Solar Energy, 193, 139–147. https://doi.org/10.1016/j.solener.2019.09.018
605 Mohapatra, T. K., & Sahu, B. K. (2018). Design and implementation of SSA based fractional order
606 PID controller for automatic generation control of a multi-area, multi-source
607 interconnected power system. 2018 Technologies for Smart-City Energy Security and
608 Power (ICSESP), 1–6. https://doi.org/10.1109/ICSESP.2018.8376697
609 Nguyen, H. N. T., Zhang, C., & Zhang, J. (2016). Dynamic Demand Control of Electric Vehicles to
610 Support Power Grid With High Penetration Level of Renewable Energy. IEEE Transactions
611 on Transportation Electrification, 2(1), 66–75. https://doi.org/10.1109/TTE.2016.2519821
612 Nwaigwe, K. N., Mutabilwa, P., & Dintwa, E. (2019). An overview of solar power (PV systems)
613 integration into electricity grids. Materials Science for Energy Technologies, 2(3), 629–
614 633. https://doi.org/10.1016/j.mset.2019.07.002
615 Puukko, J., Nousiainen, L., & Suntio, T. (2012). Three-phase photovoltaic inverter small-signal
616 modelling and model verification. 6th IET International Conference on Power Electronics,
617 Machines and Drives (PEMD 2012), 1–6. https://doi.org/10.1049/cp.2012.0351
618 Quevedo, P. M. de, Muñoz-Delgado, G., & Contreras, J. (2017). Impact of Electric Vehicles on the
619 Expansion Planning of Distribution Systems considering Renewable Energy, Storage and
620 Charging Stations. IEEE Transactions on Smart Grid, PP(99), 1–1.
621 https://doi.org/10.1109/TSG.2017.2752303
622 Quintero, G. M., Reddy Challapuram, Y., Bilbao, A., Bayne, S. B., Subburaj, A. S., & Harral, M. A.
623 (2018). Micro-grid System Modeling Efforts using PQ-Control for Single-phase and Three-

31 | P a g e
624 phase Inverter. 2018 IEEE International Telecommunications Energy Conference (IN℡

625 EC), 1–5. https://doi.org/10.1109/INTLEC.2018.8612417

626 Rahbari-Asr, N., Chow, M. Y., Chen, J., & Deng, R. (2016). Distributed Real-Time Pricing Control
627 for Large-Scale Unidirectional V2G With Multiple Energy Suppliers. IEEE Transactions on
628 Industrial Informatics, 12(5), 1953–1962. https://doi.org/10.1109/TII.2016.2569584
629 Ropp, M. E. (2009). Similarities between vehicle-to-grid interfaces and photovoltaic systems.
630 2009 IEEE Vehicle Power and Propulsion Conference, 1221–1225.
631 https://doi.org/10.1109/VPPC.2009.5289711
632 Sahoo, S., Prakash, S., & Mishra, S. (2017). Handshaking V2G strategy for grid connected PV
633 assisted charging station. IET Renewable Power Generation, 11(11), 1410–1417.
634 https://doi.org/10.1049/iet-rpg.2016.0853
635 Sakthi, N., & Kowsalya, M. (2017). Power quality improvement features on photovoltaic grid
636 connected dual voltage source inverter. 2017 International Conference on Communication
637 and Signal Processing (ICCSP), 1744–1748. https://doi.org/10.1109/ICCSP.2017.8286691
638 Syed, M. A., & Syed, R. (2019). Weighted Salp Swarm Algorithm and its applications towards
639 optimal sensor deployment. Journal of King Saud University - Computer and Information
640 Sciences. https://doi.org/10.1016/j.jksuci.2019.07.005
641 Thao, N. G. M., & Uchida, K. (2013). Control the active and reactive powers of three-phase grid-
642 connected photovoltaic inverters using feedback linearization and fuzzy logic. 2013
643 Australian Control Conference, 133–140. https://doi.org/10.1109/AUCC.2013.6697261
644 Tomić, J., & Kempton, W. (2007). Using fleets of electric-drive vehicles for grid support. Journal
645 of Power Sources, 168(2), 459–468. https://doi.org/10.1016/j.jpowsour.2007.03.010
646 Ugirumurera, J., & Haas, Z. J. (2017). Optimal Capacity Sizing for Completely Green Charging
647 Systems for Electric Vehicles. IEEE Transactions on Transportation Electrification, 3(3),
648 565–577. https://doi.org/10.1109/TTE.2017.2713098
649 Vennelaganti, S. G., Mishra, M. K., & Teja, V. R. (2016). Adaptable voltage source inverter for grid
650 integration of renewables with enhanced power quality capabilities. 2016 IEEE
651 International Conference on Industrial Technology (ICIT), 1242–1247.
652 https://doi.org/10.1109/ICIT.2016.7474932
653 Wang, R., Wang, P., & Xiao, G. (2016). Two-Stage Mechanism for Massive Electric Vehicle
654 Charging Involving Renewable Energy. IEEE Transactions on Vehicular Technology, 65(6),
655 4159–4171. https://doi.org/10.1109/TVT.2016.2523256

32 | P a g e
656 Xiao, H., Huimei, Y., Chen, W., & Hongjun, L. (2014). A survey of influence of electrics vehicle
657 charging on power grid. 2014 9th IEEE Conference on Industrial Electronics and
658 Applications, 121–126. https://doi.org/10.1109/ICIEA.2014.6931143
659 Yan, X., Sun, J., Li, Y., Qi, J., & Pan, Y. (2009). Self-Adaptive Tuning of Fuzzy PID control of PV
660 Grid-Connected Inverter. 2009 Sixth International Conference on Fuzzy Systems and
661 Knowledge Discovery, 4, 160–162. https://doi.org/10.1109/FSKD.2009.647
662 Yilmaz, M., & Krein, P. T. (2013). Review of the Impact of Vehicle-to-Grid Technologies on
663 Distribution Systems and Utility Interfaces. IEEE Transactions on Power Electronics,
664 28(12), 5673–5689. https://doi.org/10.1109/TPEL.2012.2227500
665

33 | P a g e