2014 5th IEEE PES Innovative Smart Grid Technologies Europe (ISGT Europe), October 12-15, Istanbul 1
Battery Energy Storage System Size Optimization in Microgrid using Particle Swarm Optimization Thongchart Kerdphol, Yaser Qudaih, Yasunori Mitani Department of Electrical and Electronics Engineering, Kyushu Institute of Technology, 1-1 Sensui-cho, Tobata-ku, Kitakyushu-shi, Fukuoka, 804-8550, Japan E-mail:
[email protected],
[email protected],
[email protected]
Abstract— In the recent decades, storage systems played more significant role than ever. The reason behind this phenomenon could be associated with the increase dependency on renewable energy resources in power systems. Frequency regulation is one of the important issues that Battery Energy Storage System (BESS) participated in. Furthermore, load shedding is one of frequency control methods during stand-alone operation, and the performance of frequency control improves in combination with BESS. This article proposes the feasibility of using optimal BESS with load shedding scheme when disconnection of the microgrid from the grid occurs. In this paper, Particle Swarm Optimization (PSO) is developed and presented to determine the optimal size of BESS with load shedding scheme. Results show that the optimal size of BESS-based PSO with load shedding scheme can achieve higher performance of frequency control compared to the optimal size of BESS-based analytic algorithm with load shedding scheme. The optimal sizing method of BESS with load shedding scheme is achieved by using a tool for analysis of industrial, utility and commercial electrical power system called DPL script (DIgSILENT Programming Language). Index Terms-- Battery energy storage system, Frequency control, Load shedding scheme, Particle swarm optimization.
I.
INTRODUCTION
Rapid decrease of fossil fuel resources, increase of electricity demand and environmental concerns associated with conventional generators call for an amendment of alternative electric energy generation method worldwide. Microgrid system is being considered as one of the solutions to this energy concern, and is gaining more attention recently. A microgrid is a principle concept integrating distributed generation sources (DGs) [1]. It can be viewed as a group of distributed generation sources connected to the loads in which the DGs can be fed to the loads alone or be fed to the utility grid [2]. As the output characteristics of these DGs are quite different from the conventional energy sources, the system should be capable of handling unexpected fluctuation and maintaining system reliability. Moreover, a microgrid can operate in either stand-alone or grid-connection mode. It is capable to work in parallel with the utility grid, with the capacity to switch to stand-alone operation in case of emergency situation in the grid. When an islanding operation occurs where a DG or a group of DG sources continue to
supply the microgrid system that has been separated from the utility grid, the system needs to have a master generator which can provide voltage and frequency support. Generally, a synchronous generator can fulfil this demand. When there is no synchronous generator, Voltage Source Converter (VSC) interfaced batteries can act as the master control. Presently, a battery system with the capacity of 2 MW has already been utilized to support a microgrid [3]. Battery Energy Storage System (BESS) can be used in various aspects of the power system. BESS is presently used as one key factor for sustainable energy in many countries particularly in Europe, America and Japan. Advantages of BESS include improvement of the system frequency especially when BESS is used for system frequency control. In the case of small disturbance, BESS is discharging when the system frequency is lower than 50 Hz. On the other hand, BESS is charging when the system frequency is higher than 50 Hz. In the case of large disturbance, BESS can enhance the performance of the system frequency control by integrating BESS with under frequency load shedding scheme, or under frequency generation trip, and over frequency generation trip. With these different functions, BESS can offer a good solution. It can be concluded that BESS is a rapid and flexible element for the power system [4-6]. Moreover, the purpose of optimal BESS is to smoothen the power in a system with wind or solar energy. In such a system, BESS can play a role by absorbing the surplus power and compensating the power shortage due to the uncertainties of renewable energy. However, installation of large or inappropriate size of BESS can cause frequency problems to the system and may increase the cost of the system. For these reasons, optimal sizing of BESS is an essential factor for a system [7-9]. For stand-alone operation, many optimal sizing methods have been already studied including single-objective [10] and multi-objective approach [11], [12]. This article is based on multi-objective approach. The objective of this article is to enhance frequency control and reduce an operating cost by integrating load shedding scheme with optimal sizing of BESS. Furthermore, the optimization operation can be achieved by means of linear programming method, genetic algorithm (GA) and particle swarm optimization (PSO).
978-1-4799-7720-8/14/$31.00 ©2014 IEEE
2
Particle swarm optimization (PSO) is the famous optimization technique created by J. Kennedy and R. Eberhart in 1995. They got the inspiration by social behavior of bird crowding. The PSO technique is supposed to have lower parameters than a basic genetic algorithm [13]. Moreover, It is most resistive to stop at local optimization points. Advantages of PSO include simplicity, ease of use, high convergence rate and minimal storage requirement. Also, it is less dependent on the set of the initial points compared to other methods which implies that convergence algorithm is robust [14]. With these reasons, this paper selected PSO as the proposed algorithm. PSO techniques have been already implemented in number of engineering problems. Many researches are still in progress for proving the potential of PSO in solving complex power system problems. A. Esmin et al. have proposed PSO technique which is used as a tool for loss reduction calculation and determines the required amount of shunt reactive power compensation [15]. Z. Gaing has presented PSO method which is developed to find parameters of PID controller in Automatic Voltage Regulator (AVR) system [16]. M. Bashir and J. Sadeh have presented the optimal size of BESS using PSO technique so as to reduce the annualized cost of the generation system [17]. A. Hameed et al. have presented PSO method which is designed for minimizing the size of BESS, wind turbine and solar system in order to decrease the operating cost of the system [18]. However, these papers did not consider the stability issues such as system frequency and voltage. Therefore, this paper proposes and investigates the optimal sizing of BESS using PSO method considering the stability issues. This article proposes multi-objective PSO approach which can improve fast, smooth and safe frequency control and decrease operating costs of the microgrid by integrating load shedding with optimal sizing of BESS from an emergency circumstance to a normal state in the microgrid. The proposed methodology is based on PSO algorithm in order to find the optimal size of BESS. For the performance verification, the optimal size of BESS using analytic method is selected to be a candidate for comparison against the proposed algorithm. II.
4 MVA
Bus8
Bus7
Bus3 0.2/22 kV
Bus2
Bus12
Bus11
30 MVA 150/22 kV
Grid
Control Scheme
3Ø
From the schematic structure of BESS, The output of DC voltage is drawn as [19]:
3 MVA 0.4/22 kV
Photovoltaic 3 MW
PBESS Et
6 6
'Edo
(1)
Et
S
rb 1
Ebt
cosD
6 X co
I BESSS
S
rbt
Converter
rbs
cb1
rbp cbp
+E b1
+
Ebocc -
Battery
The equivalent circuit of BESS is demonstrated as a converter connected to an equivalent battery as shown in Figure. 3. The terminal voltage of the equivalent battery is get from:
Ebt
Edo cos D Rc I BESS 3 6
Inverter
Converter
Bus5
2 MVA 6/22 kV
S
where Et is AC voltage between the line to neutral.
S
BESS
Bus6
6 6
Edo
Bus9
Bus4
Converter
Figure. 2. Schematic structure of BESS
22 kV
Bus1
BESS Transformer
Figure. 3. Equivalent circuit of BESS
L oa d1 Loa d2 Loa d3 L oa d4 3 MVA 6/22 kV Bus13
B. Battery Energy Storage System (BESS) Generally, BESS can be utilized in several implementations such as peak shaving, real power control and load leveling. This paper uses BESS for enhancing the performance of frequency control as BESS can provide active power compensation in a short period of time. This strategy can be considered as a new method. Recently, a huge number of electric companies and independent system operators have shown growing interest in BESS due to decreasing cost of batteries. With the fast development of technologies, BESS is expected to be used in several applications [6] including the one proposed in this paper. The structure of BESS consists of power converters, battery cells and control parts which are shown in Figure. 2.
SYSTEM CONFIGURATION
A. Microgrid System
Hydro Gen 2 MW
connected to the utility grid. Figure. 1 shows the microgrid system which consists of a 1.2 MW mini-hydro generator, 2 MW hydro generator and 3 MW photovoltaic sources. BESS is connected to the microgrid system at the bus 1. The system also consists of a group of feeders which could be part of the distribution system. There are some critical loads which require a local generation and the non-critical load feeders are not connected to any local generation.
Load 5
Bus10
6
S
X co I BESS
(2)
According to the equivalent circuit of BESS, the expression of DC current flowing into the battery can be expressed as:
I BESS
Figure. 1. Microgrid architecture
A typical characteristic of a microgrid is that it can be operated either in grid-connected operation or stand-alone operation. Under normal operation, the microgrid is
Et cos D1 cos D 2
where,
Ebt Eboc Eb1 rbt rbs rbp Eboc I BESS 1 STbp
(3) (4)
3
Eb1
rb1 I BESS 1 STb1
(5)
Tbp
rbpCbp
(6)
Tb1
rb1Cb1
(7)
From the converter circuit analysis, the active and reactive power is absorbed by BESS as [19]: 3 6 (8) PBESS Et I BESS cos D 1 cos D 2
S
3 6
QBESS
S
Et I BESS sin D 1 sin D 2
(9)
Only incremental active power is considered in load frequency control. Therefore, P-modulation strategy is selected to this paper [6]. For P-modulation, α1= - α2= α.
PBESS QBESS
6 6
S
(10)
0
Kb 'f 1 STb
(13)
The ∆f is the beneficial feedback from the power system in order to provide a damping effect. where αi is the firing delay angle of converter i. Ed0 is the maximum DC voltage of the batteries. Eb1 is the battery overvoltage. Ebt is the terminal of equivalent battery. Eboc is the battery open circuit voltage. IBESS is the DC current though the battery. PBESS is the active power provided by the batteries. rbt is the connecting resistance. rbs is the battery internal resistance. rbp is the self discharge resistance. rb1 is the overvoltage resistance. ∆f is the frequency deviation. Xco is the commutating reactance. ∆Eco is the DC voltage without overlap. Kb is the control loop gain. Tb is the measurement device time constant. From linearizing equations (1) to (13), the block diagram of BESS can be shown in Figure 4. K b 1 ST
'E d +
b
6
'E p Edo cosD +
0 I BESS
6 X co
S
' E co
' PBESS
-
+
3
6
+
'Ebt
6
6
+
-
Eb1 +
E boc
1 rbs rbp
rb1 1 STb1
'I BESS +
6
+
rbp 1 STbp
I
A. Structure of Developed Algorithm DIgSILENT PowerFactory is a computer aided engineering software for the analysis of industrial, utility and commercial electrical power system. This program provides the programming language called DIgSILENT Programming Language (DPL) used for developing automated scripts which connects other objects or elements in DIgSILENT PowerFactory. The structure of this language is similar to C++ language. Figure. 5 shows the structure of the DPL script used in developing the optimal sizing of BESS [20]. Resulting parameters
DPL Internal Varibles
Database Input parameters ComLFD
(11)
(12) 'PBESS I 'Ed Then, the use of BESS in load frequency control is reached by a damping signal ∆Ed.
'f
PROPOSED CONTROL METHOD
Internal Object
Et I BESS cos D
0 BESS
'E d
III.
0 BESS
Figure. 4. Linearized BESS model for load frequency control
Sub 1 Sub S 2 Sub S 3
SetFilt
External Object
Figure. 5. Structure of DPL script for optimal sizing of BESS with load shedding scheme
B. Optimal Sizing of BESS-based PSO Algorithm with Load Shedding Scheme Particle swarm is an algorithm technique for searching optimal parameters of complicate search spaces. PSO is started with a group of random particles and finds for optimal parameters by updating generations. In each iteration, each particle is updated by two values. These values are called Pbest and Gbest consequently. Pbest is the best solution acquired by each particle itself in all of the previous generations. Gbest is the best value obtained by any particle in all previous iterations. This value is called the best global solution [21]. Each particle updates its position and velocity by using equation (14) and (15). vi 1 vi c1r1 Pbest xi c2 r2 Gbest xi (14)
xi 1 vi 1 xi (15) where ith is the iteration number. jth is the particle number. vi is the velocity of a particle at iteration ith. xi is the position of a particle at iteration ith. Pbest is the best solution at iteration ith. Gbest is the best global solution at iteration ith. r1 is the random number one between 0 and 1. r2 is the random number two between 0 and 1. c1, c2 are the learning factors The learning factors have significant effects on the algorithm convergence rate. Further information for PSO can be found in [13]. The PSO parameters chosen for this proposed algorithm are: •The number of particles in swarm (NP) = 20 •The number of iterations (NI) = 30 • Learning factors are C1= C2 = 1.49445 • Inertia weight (w) = 0.7920 •The number of dimension (D) = 1
4
The main steps for the proposed optimal BESS-based PSO algorithm are: min
max
Step 1) Initialize the parameters ( PBESS , PBESS , C1, C2, w,
Step 2) Step 3) Step 4)
Step 5) Step 6) Step 7) Step 8)
NI, NP) and iteration i =1 with random position (xj) and velocity (vj). Load shedding is set based on the measured power imported by the grid at the moment of islanding. Start the particle j=1 in the swarm. Execute the objective function for particle jth in iteration ith. Find Pbest and Gbest for particle jth in iteration ith. If the fitness value (xj) ≤ Pbest, the program sets Pbest = xj. Increase particle jth by 1. Then, check the condition If particle jth +1≤ NP, go back to step 3). Calculate and adjust the new position (xjnew) and velocity (vjnew) for particle jth in iteration ith. Increase iteration ith by 1. Then, check the condition if iteration ith +1≤ NI, go back to step 2). If iteration ith +1> the number of iteration (NI), the process ends and optimal size of BESS = Gbest.
The overall flow chart of the optimal sizing of BESSbased PSO with load shedding is shown in Figure 6. Start
C.
Optimal Sizing of BESS-based Analytic Algorithm with Load Shedding Scheme The overall flow chart of the optimal sizing of BESSbased analytic algorithm with load shedding is shown in Figure 7. The solution algorithm of this method is as follows: Step 1) Read the input data of the microgrid system in DIgSILENT PowerFactory program and get the power values of photovoltaic (PPV), hydro generator (PHV), mini hydro generator (PMini) and the load 1 to 5 (PL). Step 2) Set the desired power of load shedding (PS). Load shedding is based on the measured power imported by the grid at the moment of islanding. Step 3) Calculate the total power of loads (PLT) after using the load shedding. Then, the program calculates the differential power (PDP) between generations and loads. Step 4) Set the differential power (PDP) equal to the power of BESS (PB). Step 5) Check the condition if the system frequency is within the values set by the user. If the system frequency does not satisfy the condition, the program will increase PBESS by 0.1 MW and the condition is run again. This step will continue until it meets the condition. Step 6) End of process.
Initialize swarm min max Parameters NI, NT ,w, C1, C2,PB ,PB PBESS PBESS Swarm population: x, v, Gbest, Pbest with random states and probabilities
Start Initialize the output power; Parameters PPV, PHV, PMini, PL
Iteration i = 1
Set PS
Particle j = 1 Execute the fitness value of particle jth (xj) and calculate size of BESS, frequency of microgrid
Yes
Yes Calculate the total power of load : PLT P L PS
If fitness value (xj) < Best global fitness (Gbest)
Calculate the differential power :
No Set Pbest = fitness value (xj)
PDP PLT PPV PHY PMini
Next particle: j = j+1
Check: j ≤ NT
No
PS < PL
Set PBESS
No
PDP
Yes Adjust Gbest(i) = the best of Pbest (i)
Next iteration: i = i+1
Check: i < NI
PB t 0 Update velocity and position with equation (14) and (15)
Yes
No End of process; Optimal size of BESS = Gbest
Figure. 6. Overall flow chart of proposed optimal sizing of BESS-based PSO algorithm with load shedding scheme
No
Yes Calculate : Size of BESS, Frequency of Microgrid Update size of BESS:
PBESS PBESS0.1
No
Check
F Min d F d F Max
Yes End
Figure. 7. Overall flow chart of the optimal sizing of BESS-based analytic algorithm with load shedding scheme
5
D.
Objective Function To minimize the power of BESS, the final objective function is chosen and expressed as: Minimize f = min (PBESS)
However, no difference is observed after 20 th iterations. The reason for this is that this algorithm is symmetrical around the midpoint.
(16)
under the constraints of: min max Fmin d F d Fmax and PBESS d PBESS d PBESS
where F = Frequency of the microgrid system (Hz). PBESS = Active power of BESS (MW). IV.
SIMULATION RESULTS
If the generation cannot match the load demand in a short time during a large disturbance such as three phase fault, the system frequency will drop drastically. This significant drop will likely cause further extreme consequence such as system collapse. In this case, under frequency load shedding scheme plays an important role in frequency control and BESS can improve the performance of frequency control in load shedding scheme. In this part, the optimal sizing of BESS with load shedding is accomplished by using DPL script in DIgSILENT PowerFactory, and the performance of the system frequency is discussed in the simulation results. For the performance verification, the optimal sizing of BESS-based analytic algorithm is selected to be the candidate for comparison against the proposed algorithm. There are three power plants namely Mini-hydro, Hydro, and Photovoltaic with average output power of 1.2 MW, 2 MW and 3 MW respectively. Moreover, there are two critical loads with power of 1.85 MW and 1.9 MW (Load 1, Load 4) and three non-critical loads with power of 1.7 MW, 1.75 MW and 2.4 MW (Load 2, Load 3, Load 5) at the moment of islanding. A three phase fault has occurred on bus 7 at t = 10.0s and the system has applied the load shedding scheme to the non-critical load 5. Normally, the time to clear the fault and restore the system after the islanding operation occurred takes around 1 hour. Thus, this paper assumes that BESS can support the microgrid system for about 1 hour after the system is isolated from the main grid.
Figure. 9. Improvement of the system frequency with load shedding
Figure. 10. Improvement of the system voltage with load shedding
In a case where the system cannot apply load shedding scheme when a three phase fault occurs at bus 7, the system will collapse as shown in Figure 9 and 10.
4.5
4
Fitness value
3.5
3
Figure. 11. Frequency deviation of the microgrid system after islanding
2.5
2
1.5
1
0
5
10
15
20
25
30
Iteration
Figure. 8. Convergence rate of the optimal sizing of BESS-based PSO
Figure. 8 displays the convergence rate of the optimal sizing of BESS-based PSO. The PSO algorithm is converged to its final state approximately after 15 iterations. It is evident that attainment is achieved with Gbest after 15th iterations. The slightly attainment is reached with Gbest after 20th iterations.
Due to the three phase fault, load shedding is a method of frequency control protection. In this paper, BESS and load shedding scheme work together at the same time. The performance of the system frequency with optimal size of BESS and load shedding scheme is observed and shown in Figure. 11. From this figure, the magnitude of the frequency deviation with the optimal sizing of BESS-based analytic algorithm is higher and takes more time to stabilize than the optimal sizing of BESS-based PSO. However, without optimal size of BESS, the magnitude of the frequency deviation reaches 51 Hz and takes longer to stabilize. In case
6
of no BESS, the system frequency dropped drastically because the power supply cannot meet the load demand. It can be seen in Table I that in case of the optimal sizing of BESS, the performance of system frequency is much better when the optimal sizing of BESS-based PSO is applied simultaneously with load shedding scheme. The influence of BESS is the essential point in this paper and df/dt method is not included.
BESS with load shedding was clearly illustrated and approved. REFERENCES [1]
[2]
[3]
[4]
[5]
[6]
[7] Figure. 12. Voltage deviation of the microgrid system after islanding
Besides, the voltage deviation of the microgrid system is shown in Figure 12. From this figure, the optimal sizing of BESS-based PSO can still maintain the voltage deviation better than the optimal sizing of BESS-based analytic algorithm. It is clear that the microgrid system with optimal size of BESS-based PSO is more robust and stable than the others. TABLE I FREQUENCY AND VOLTAGE OF MICROGRID SYSTEM FOR EACH CASE AFTER ISLANDING OPERATION BESS Frequency (Hz.) Voltage (pu.) Case (MW) Reference : 50.00 Reference : 0.9890 No BESS
0
31.0258
0.8405
BESS without optimal size BESS-based PSO
2.0000
50.3011
0.9950
1.3124
50.0012
0.9890
BESS-based Analytic
1.4000
50.1905
0.9921
[8] [9]
[10]
[11]
[12]
[13]
[14]
[15]
V.
CONCLUSION
This paper proposes a control scheme using the optimal sizing of BESS-based PSO with load shedding scheme in order to improve the system frequency after islanding occurred. Based on the results, BESS is shown a good performance in load frequency control and can be used for emergency control purposes when the system load shedding is unavoidable. Results show that the optimal sizing of BESSbased PSO can gives better stability than the optimal sizing of BESS-based analytic algorithm. Moreover, the optimal sizing of BESS-based PSO can also provide the sustainable voltage to the system much better compared to the optimal sizing of BESS-based analytic algorithm. It is obvious that the proposed algorithm provides quicker, smoother and more secure system stability than the optimal sizing of BESS-based analytic algorithm from the emergency situation to the normal equilibrium state. In addition, the usage of optimal
[16]
[17]
[18]
[19] [20] [21]
H. Karimi, H. Nikkhajoei, and R. Iravani, “Control of an electronically coupled distributed resource unit subsequent to an islanding event,” IEEE Trans. Power Delivery, vol. 23, no. 1, pp. 293-501, January 2008. R. Wai and W. Wang, “Grid-connected photovoltaic generation system,” IEEE. Trans. Circuits System, vol. 55, no. 3, pp. 953-964, April 2008. F. Katiraei, M. Iravani, and P. Lehn, “Micro-grid autonomous operation during and subsequent to islanding process,” IEEE Trans. Power Delivery, vol. 20, no. 1, pp. 248-257, January 2005. P. Mercier, R. Cherkaoui, and A. Oudalov, “Optimizing a battery energy storage system for frequency control application in an isolated power system,” IEEE Trans. Power System, vol. 24, pp. 1469-1477, August 2009. A. Oudalov, D. Chartouni, and C. Ohler, “Optimizing a battery energy storage system for primary frequency control,” IEEE Trans. Power System, vol. 22, pp. 1259-1266, August 2007. S. K. Aditya and D. Das, “Application of battery energy storage system to load frequency control of an islanded power system,” International Journal of Energy Research, vol. 23, pp. 247-258, 1999. Y. Halevi and D. Kottick, “Optimization of load shedding system,” IEEE Trans. Energy Conversion, vol. 8, pp. 207-213, June 1993. P. M. Anderson and M. Mirheydar, “A Low-order frequency response model,” IEEE Trans. Power System, vol. 5, pp. 720-729, August 1990. P. M. Anderson and M. Mirheydar, “An adaptive method for setting underfrequency load shedding relays,” IEEE Trans. On Power System, vol. 7, pp. 647-655, May 1992. W. Kellogg, M. Nehrir, G. Venkataramanan and V. Gerez, “Generation unit sizing and cost analysis for standard-alone wind, photovoltaic, and hybrid wind/PV systems,” IEEE Trans. Energy Conversion, vol. 13, pp. 70-75, March 1998. B. Borowy and Z. Salameh, “Methodology for optimally sizing the combination of a battery bank and PV array in a wind/PV hybrid system,” IEEE Trans. Energy Conversion, vol. 11, pp. 367-375, June 1996. G. Shrestha and L. Goel, “A study on optimal sizing of stand-alone photovoltaic stations,” IEEE Trans. Energy Conversion, vol. 13, pp. 373-378, December 1998. J. Kennedy and R. Eberhart, “Particle swarm optimization”, in Proc IEEE Int. Conf. Neural Networks, Perth, Australia, vol. 4, pp. 19421948, 1995. Z. L. Gaing, “Particle swarm optimization to solving the economic dispatch considering the generator constraints”, IEEE Trans. Power Systems., vol. 18, no. 3, pp. 1187-1195, August, 2003. A. A. Esmin, G. Lambert-Torres, and A. C. Zambroni de Souza, “A hybrid particle swarm optimization applied to loss power minimization”, IEEE Trans. Power System, vol. 20, no. 2, pp. 859-866, May 2005. Z. L. Gaing, “A particle swarm optimization approach for optimum design of PID controller in AVR system”, IEEE Trans. Energy Conversion, vol. 19, no. 2, pp. 284-391, June, 2004. M. Bashir and J. Sadeh, “Size optimization of new hybrid stand-alone renewable energy system considering a reliability index”, IEEE Int. Conf. Environment and Electrical Engineering, pp. 989-994, May, 2012. A. Hameed, M. Elhagri, and A. Shaltout, “Optimum sizing of hybrid WT/PV system via open-space particle swarm optimization”, Iranian. Conf. Renewable Energy and Distributed Generation, pp. 50-60, March, 2012. E. Kimbark, “Direction current transmission”, NY: Wiley, vol. 1, 1971. Version 14.1 PowerFactory, User’s Manual, 1st ed. Gomaringen, Germany: DigSILENT, vol. 2, 2011. T. Zhou and W. Sun, “Optimization of battery-supercapacitor hybrid energy storage station in wind/solar generation system”, IEEE Trans. Sustainable Energy. Vol. 5, no. 2, pp. 408-415, April, 2014.
