M. Karakse 1 K. Murat 1 1 Firat University Computer Engineering Elazig, Turkey mkarakose@firat.edu.tr kmurat@firat.edu.tr E. Akn 1 K. S. Parlak 2
2 Firat University Vocational School of Technical Sciences Elazig, Turkey eakin@firat.edu.tr ksparlak@firat.edu.tr
Abstract The purpose of reconfiguration process in PV systems is to find the connection structure of the highest amount of power obtained by the system under partial shading conditions. In this paper, a new genetic-algorithm based approach was presented for the reconfiguration. The system tries to maximize the obtained power in the same sub-module combining the ones that have the most approximate radiation value to each other between the panels exposing to partial shading. The proposed method only needs to short-circuit currents pertaining to an adaptive and fixed panel as the input parameter. So, it has a flexible structure that can work independently from the hardware features of PV panels. It creates the initial population generating random values by benefiting from short-circuit current information it receives. Based on these values, optimum connection diagram can be obtained in a short time. The most important advantage of the proposed algorithm is its applicability in PV systems including a great number of panels and the efficient results it generated in terms of the operating speed. For the test of the proposed method, test data were obtained by the help of simulation prepared in MATLAB-SIMULINK environment. Then, these received data were used in genetic algorithm based reconfiguration algorithm, and the required result was obtained. The obtained results revealed that the proposed approach yielded more efficient results in PV systems. Keywords Reconfiguration, genetic algorithm, PV systems, optimization. I. INTRODUCTION Partial shading in PV systems is an important factor affecting the productivity of system. The decrease at the amount of radiation falling upon the panels reduces the output energy, as well [1]. There are two basic solutions in application to draw more current from the system. One of these is maximum power point tracking (MPPT), and the other is reconfiguration [2]. MPPT is a method that aims operating of the system at a maximum power point under any circumstances. There have been several studies related to this subject in the literature [3-10]. The basic approach in MPPT techniques is the follow- up of the maximum point at P-V curve. By this means, it is aimed to obtain constant maximum power from PV panels. However, the applied MPPT techniques are special to the system. Namely, it should be redesigned for each system that will be applied. Moreover, these techniques depend upon the connection types of PV series. That is, MPPT techniques try to find the maximum power that can be obtained from the connection at that time. Moreover, the system can have a connection structure that can obtain higher power rather than the shading conditions. Reconfiguration method is used to overcome these disadvantages [11]. The process of reconfiguration aims to obtain maximum power values changing the structure of the system in PV series dynamically [12]. In the literature, there have been various reconfiguration methods, and these techniques are inefficient in big-size PV series. In their study Cheng et al. [13] proposed a method that used sub-module stresses and open-circuit current of PV cells. As presented in Fig. 1, the method included a fixed panel, adaptive panel, and switching matrix. The study was tested using a 3x10 PV panel, and the advantages of blurred control were told about. In another study [14], Rough Sets Theory (RST) was applied for the reconfiguration, and an automatic reconfiguration system was proposed. A 3x2 PV series was used for the test process. In some approaches, switching was used upon the electrical model of PV cells. In such kind of studies, because the processes were re-performed for each PV, these become disadvantageous in terms of both time and process complexity.
C o n t r o l
A l g o r i t h m Fixed Panel Adaptive Panel I S w i t c h i n g
M a t r i x
C i r c u i t R V 1 , V 2 , , V m I I I I output V 1 , V 2 , , V m
Fig. 1. Reconfiguration approach in [13]. 978-1-4799-2399-1/14/$31.00 2014 IEEE 23 In their study, Velasco et al. [15] used Electric Array Reconfiguration (EAR) strategy, and proposed a method presenting a mathematical method. In the study, it was aimed to minimize the difference between the maximum and minimum radiance levels in a PV series, and 6 PV modules were tested. In a similar study, El-Dein et al. [16] used a bipolar m-throws switching circuit for the reconfiguration process, and presented a global mathematical model. They actualized the test process in a 6x4 series. Some of the methods presented for the fixed PV systems are designed through receiving the catalogue information or some other input information of PV series, and applicability of these methods do not have the adequate flexibility. The method either depends upon the system that has the catalogue information it uses or expects the calculation of values measurement of which takes extra time. In the study of Cheng [17], a model-based reconfiguration algorithm was presented. The method requires the calculation of shading degree using the photo-generated current of each sub-module. The basic function of the reconfiguration method is its having a connection type that will provide the maximum power. In the literature, there are various algorithms that help finding the best among the possible connection types. However, these approaches are the methods that are appropriate to be used in small-sized series and extend the process of finding a result in big-sized series. In their study, Nyugen et al. [18] presented a method that used the bubble sort algorithm for the decision process. Although they performed the test process in a 3x3 small-sized series, this does not seem appropriate to a real-time study in big-sized series. Another presented method [19], used bubble sort algorithm to determine the optimum array connection. But it shows the test results for 3x3 series. This doesnt seem appropriate to apply a real time system. In this study, a new genetic algorithm-based reconfiguration approach was presented. The method prepared by using adaptive and fixed panels has the ability to operate as real-time in PV systems at any sizes with the developed reconfiguration algorithm. Especially in genetic based reconfiguration algorithm section, it was provided to calculate the possible connection types, then to determine the connection series that will give the maximum power based on the random values created by the system instead of eliminating. By this means, the system finds the connection structure with the maximum energy much faster, and responds to varying partial shading conditions much quicker. Furthermore, the presented method has a flexible structure operating independently from the system features. As the input parameters, it only receives short-circuit currents pertaining only to adaptive and fixed panels. Because it does not require any catalogue information, voltage or temperature information except from this information, it provides direct applicability upon different systems. Under partial shading conditions, the effects of the proposed method were tested in simulations at the end of the study. The obtained results approved the advantages of the method.
II. RECONFIGURATION FOR PV ARRAYS
A PV cell has the ability to absorb the solar energy and convert this energy into electric energy via the photons [20]. This obtained energy can be usable so long as the radiation over the cell continues. Moreover, formation of partial shading over the cells causes the productivity to decrease under both individual and simultaneous operating conditions in these cells [21]. In order to affect at a minimum level from this negative situation, PV cells can be connected in series or parallels while operating. However, only these connection types can be inadequate in module-based studies. For that reason, three different connection types are used for PV modules. These connection types presented in Fig. 2 are S&P, TCT, and BL type connections. But, these connection types also vary in terms of productivity [16, 22]. Reconfiguration is one of the methods used to overcome negative situation caused by the partial shading in PV series operating within the scope of a system. In this method, the system is predicated to rearrange itself according to varying radiation. By this means, a dynamic operation is provided instead of a fixed connection structure. It is aimed to provide the maximum power that can be obtained from the series under any circumstances. The system includes three basic parts in a reconfiguration process that will be actualized in a PV system. These parts are fixed panel, adaptive panel, and switching matrix circuit [2]. The fixed panel is the main operating part of the system that has more PV cells. During the process, it does not undergo a change and is fixed. Each line on this panel creates the sub- modules, and it included mxn number of panels (m number of sub-modules including n number of cells) (F 1 , F 2 ,, F m ). The second part is adaptive panel. This part includes m number of panels connected to each sub-module (A 1 , A 2 ,, A m ). Under uniform radiation conditions, each panel in the adaptive panel connects to a sub-module over the fixed panel. The third basic part is switching matrix circuit providing the connection between the adaptive and fixed panels. Each switch in this circuit presented in Fig. 3 is designed in a way that will connect any PV panels in adaptive panel to any sub-modules in fixed panels. Because each panel is bipolar, the total switch element number in a switching matrix circuit will be 2x(m x m).
a) b) c) Fig. 2. Array configurations for reducing the power loss under partial shading a) SP configuration b) TCT configuration c) BL configuration
24 S(1,1) S(1,2) S(1,m) S(2,1) S(m,1) A 1 A 2 A m F 1 F 2 F m
Fig. 3. Switching matrix circuit for reconfiguration
The reconfiguration process starts when the value of radiation over the system changes, namely when the partial shading occurs. The system generates the required input values according to the radiance amount over the sub-modules in the fixed panel. These values are directed to a decision-making mechanism expected to find a connection order providing the maximum power that can be obtained by the system simultaneously. The result obtained here is directed to switching matrix circuit by the help of a controller as the subsequent stage. Through this stage, the order of connection between the adaptive panel and fixed panel is established, and a new connection order to the system of sub-modules in the fixed panel is provided. III. PROPOSED APPROACH The most important process for the reconfiguration in PV series is the presence of optimum configuration order. The main difficulty here is the great number of the way of possible connections as the size of the series increase. In this study, a very efficient approach that provides the applicability of reconfiguration algorithms in real-time and big series sizes was presented. For example, in a series that have a 10 x 10 panel, determining the best of billions of possibilities is impossible or hard to apply. However, in this study it was possible to perform reconfiguration in systems under any sizes and conditions through the genetic based reconfiguration algorithm. The block diagram of the proposed approach could be seen in Fig. 4. A. PV System Structure The structure of the proposed approach includes three basic blocks. The first one is the fixed panel part connected through a TCT connection structure including m number of sub-modules. The connection structure in this part is fixed during the operation. The second part is the adaptive panel including panels as many as the number of sub-modules in the fixed panel. The third part is the switching matrix circuit connecting the adaptive and fixed panels to each other. This circuit is designed in a way in which it will connect a panel in the adaptive part to a sub-module in fixed part. The data acquired from the fixed and adaptive panels are transferred to a data acquisition unit. The acquired data are sent to control algorithm for data processing, and the obtained optimum result is directed to a switching matrix circuit by the help of a controller, and actualized on the system. S w i t c h i n g
M a t r i x
C i r c u i t Adaptive Panel Fixed Panel Data Acquisition Optimization Based Reconfiguration Algorithm Controller A 1 A 2 A 3 A m F 1 F 2 F 3 F m I (sc) A I (sc) F
Fig. 4. Block diagram of proposed reconfiguration approach
The purpose of the method proposed in this study is to determine the sub-modules and the panels that have the most proximate radiation and shading values among the ones in PV series. So, current limiting and productivity loss that can arise from the type of connection will be decreased. B. Data Acquisition When the other methods in the literature are analyzed, it can be noticed that from the system processes can be actualized calculating several input values such as temperature, voltage, and shading degree. However, because including these values into reconfiguration process after calculation takes much time and cost, it affects the performance negatively. Short circuit currents include processes that are easy to be measured and do not cause waste of time. By this means, it does not affect the performance of the system rather than the other data types. Therefore, short circuit current were used as the input parameter in study of the proposed method. Short circuit currents of the adaptive panel (I (sc) A) and short circuit currents of the fixed panel (I (sc) F) were included into the data acquisition unit.
C. Optimization Based Reconfiguration Algorithm In the reconfiguration processes in other studies, finding the possible connection orders, and choosing of the optimum ones among these take too much time. And this causes the system to respond to the partial shading conditions late and the productivity to decrease. In this study, a genetic based reconfiguration algorithm was presented. Because genetic algorithms reveal fast and efficient results, these are the methods preferred in real-time applications [23, 24]. Using the genetic algorithm here, highly beneficial results were obtained in order to find the most optimum of the possible connection configurations especially in big sizes. The flow diagram of the proposed approach was presented in Fig. 5.
Creating the beginning population: A sample connection definition showing the structure of the chromosomes creating 25
Create the beginning population by using short-circuit currents Apply the fitnnes function Crossing Mutation New generation Whether if reached maximum iteration or solution End Start Y N Selection Fig. 5. Flow diagram of genetic based algorithm
the population was presented in Table I. F indicates the fixed panel in the system, and A indicates the adaptive panel in the system.
According to the sample connection in Table I; A 1 and A 3 terminals were connected to F 1 , A 2 terminal was connected to F 2 and F 3 was left as blank. According to this, we obtain a three-bit value for each F indicating the connected terminal values with disconnected terminal values. The nine-bit expression created when we bring these values together was shown in Fig. 6(a). The beginning population pertaining to our genetic based algorithm will include the chromosomes that have the aforementioned structure. Each chromosome was mxm byte, and determined as random.
Applying the fitness funtion: The first step for the fitness value that will be calculated for each chromosome in optimization algorithm is the control of the chromosomes validity. Namely, it is to provide the elimination of the chromosomes connection structure of which are not possible from the others. For this, the values corresponding to F bits in each chromosome are added. The bits in the obtained result were multiplied. If the obtained value is 1, it is added with 0, and if different from 1, it is added with X max . X max , is the TABLE I. A SAMPLE CONNECTION BETWEEN ADAPTIVE AND FIXED PANEL Statik Ksm Adaptif Ksm F1 A1,A3 F2 A2 F3 -
maximum value a chromosome can take from this process. When the chromosome structure in Fig. 6(a) is applied to the process, the situation in Fig. 6(b) is obtained.
Short circuit currents are used for the second step in determining the genetic based reconfiguration algorithm. If I (sc) F is taken as the short circuit current pertaining to the fixed part, and I (sc) A is taken as the short circuit current of the adaptive part, then Y value is calculated by the help of Equations (1) (4).
n i I I I i A sc F sc i sc i ,... 2 , 1 , ) ( ) ( ) ( = + = (1) ] ,..., , max[ max ) ( ) ( ) ( 2 1 n sc sc sc I I I I = (2) ] ,..., , min[ min ) ( ) ( ) ( 2 1 n sc sc sc I I I I = (3) I I Y min max = (4)
Finally, fitness function should be as such:
Y X x fitness * ) ( = (5)
Selection, Crossing and Mutation: The process of selection among the chromosomes is performed as random. The crossing is actualized through the randomly chosen two chromosomes. Within each chromosome, two-point crossing process is performed with a randomly chosen index. The mutation process is actualized through the transformation of a gene within a randomly chosen chromosome.
Creating new generation: The new chromosomes obtained as result of the crossing and mutation are added to the population. Then, the three chromosomes with the biggest value of fitness function are excluded from the population.
Evaluation: Whether the required maximum iteration or repeated solution can be obtained or not is controlled, and the termination of the application is decided. A1 A2 A3 A1 A2 A3 A1 A2 A3 F1 1 0 1 F2 0 1 0 F3 0 0 0 a)
F1 F2 F3 + + = 111 1 * 1 * 1 = 1 X = 1 + 0 = 1 101 010 000 b) Fig. 6. Bits of array connection a) Chromosome structure b) Method to eliminate the connections which cannot apply 26 IV. SIMULATION RESULTS In this study, the testing of the proposed method was actualized with two different simulations. The first one was the testing process performed using the 3 x 3 size system. The second simulation was performed increasing the PV series up to 10 sub-modules. Short circuit currents used in simulations were obtained using the PV system model prepared in MATLAB-SIMULINK environment before. The testing process was actualized in a computer with Intel Core i7- 2630QM 2.00 GHz processor and 8 GB RAM. In the first simulation study, different radiation values were given to PV panels, and partial shading conditions were created. The radiation values obtained under these conditions were presented in Table II. The short circuit currents measured for these radiation values were as presented in Table III. The method proposed using these values was tested accepting the crossing process as 3 bits and mutation rate as 0.01. The way of connection obtained as the result, the period of obtaining, and the rate of obtaining were given in Table IV with different beginning populations. These results obtained through the proposed method were compared with the reconfiguration approaches used in two different studies [18, 19]. During this comparison, the input parameters of the three methods and the periods for the approaches to obtain a result were regarded as basic. The obtained comparison results were presented in Table V. In the second simulation study, the test was performed increasing the number of used PV series between 4 and 10. The period to find the optimum connection order pertaining to three different methods and the obtained results TABLE II. RADIATION VALUES OF PV SYSTEM No Adaptive Panel (W/m2) Fixed Panel (W/m2) 1 300 1,000 900 800 2 700 600 500 400 3 700 400 200 300 TABLE III. SHORT CIRCUIT CURRENT VALUES OF PV SYSTEM No Adaptive Panel (A) Fixed Panel (A) 1 1.1380 4.1740 2 2.6560 6.8300 3 2.6560 10.2400 TABLE IV. RESULTS FOR THE FIRST SUMULATION The number of chromosome in beginning population Connection Process time (ms) Ratio of true result (%) F1 F2 F3 10 000 000 111 66 22 20 000 000 111 298 45 50 000 000 111 729 89 100 000 000 111 2,539 100 300 000 000 111 21,667 100 TABLE V. COMPARISON OF BUBBLE SORT, MODEL BASED VE GENETIC BASED METHODS Reconfiguration Method Input Parameter Process time (ms) Bubble sort [19] Short circuit currents 1,478 Model-Based[18] Photo-generated currents 686 Our Genetic-Based Short circuit currents 300
were given in Table VI. The comparison of these result values was simulated through MATLAB, and the obtained graphic was shown in Fig. 7. When both simulation results were analyzed, the performance of the proposed method under partial shading conditions could be noticed. When the obtained results were analyzed in the first test process, we can say that the input parameters and the used method were efficient during the result-finding period. Especially when the results obtained for the bubble-sort method and the proposed method were analyzed, it can be noticed that although both methods took the same input parameters, there was nearly four-fold difference among the periods to obtain a result. In the second test process, we could see the difference among the three methods more clearly when the number of sub-modules in PV system was increased. Especially in 4 or more PV series, the number of possible connection types reach to fairly high numbers. So, the method used to find the most optimum among these connection types is the determining factor in response speed of the system. TABLE VI. TEST RESULTS FOR MORE PV SERIES Number of sub- modules Bubble sort[19] (ms) Model-Based[18] (ms) Our Genetic- Based (ms) 4 24,576 10,800 6,000 5 375,000 148,576 18,000 6 5,598,720 2,188,320 36,000 7 84,707,280 28,646,680 75,000 8 1,321,205,760 587,202,560 90,000 9 21,427,701,120 10,523,422,720 105,000 10 362,880,000,000 151,280,000,000 240,000
4 6 8 10 12 14 16 0 2 4 6 8 10 x 10 7 Number of sub-modules T i m e
( m s )
Bubble Sort Method Model Based Method Genetic Based Method
Fig. 7. Comparison result for more PV sub-modul. 27 When the graphic in Fig. 7 was analyzed, it could be noticed that the proposed method was more appropriate for the big-size systems rather than the other methods. V. CONCLUSIONS In this study, a new genetic algorithm based reconfiguration approach has been proposed to increase the productivity of PV systems which decreased in their rate under the partial shading conditions. The system was prepared using an adaptive panel, a switching matrix circuit, and a fixed panel connected in TCT structure. The series were exposed to partial shading using a PV system model prepared in MATLAB-SIMULINK environment, and the short circuit currents were calculated. These obtained short circuit currents were given to the developed genetic based reconfiguration algorithm, and the optimum connection order was found. In simulation studies, two different test results pertaining to the proposed method were presented. Firstly, the method was operated upon a system at a 3x3 low level, and the obtained results were experimented at a similar level, and compared with different studies. In the second simulation study, the system was tested using more numbers of sub-modules, and the comparison results were presented through different methods. When both test results were considered, it was noticed that the proposed reconfiguration method was faster than the other methods, and was more applicable in operation of real-time systems. ACKNOWLEDGMENT This study has been supported by The Scientific and Technological Research Council of Turkey (TUBITAK 1001 Programme) under Research Project No: 112E214. REFERENCES [1] H. Patel, V. Agarwal, MATLAB-Based Modeling to Study the Effects of Partial Shading on PV Array Characteristics, IEEE Transactions on Energy Conversion, vol. 23, no. 1, pp. 302-311, March 2008. [2] A. Bidram, H. Davoudi, R. S. Balog, Control and Circuit Techniques to Mitigate Partial Shading Effects in Photovoltaic Arrays, IEEE Journal of Photovoltaics, vol. 2, no. 4, pp. 532-547, October 2012. [3] R. Garraoui, B. H. Mouna, L. Sbita, MPPT Controller For a Photovoltaic Power System Based on Fuzzy Logic, 10 th International Multi-Conference on Systems, Signals & Devices (SSD) Hammamet, Tunusia, pp. 1-6, March 18-21, July 2013. [4] C. S. Chin, Y. K. Chin, B. L. Chua, A. Kiring, K. T. K. Teo, Fuzzy Logic Based MPPT for PV Array Under Partially Shaded Conditions, 2012 International Conference on Advanced Computer Science Applications and Technologies (ACSAT), pp. 133-138, 26-28 November, 2012. [5] M. Balato, D. Gallo, C. Landi, M. Luiso, M.Vitelli, Smulation and Laboratory Characterization of a Hybrid MPPT Tecnique Based on Fast Estimate of the Maximum Power Voltages in PV Application, IEEE International Conference Instrumentation and Measurement Technology Conference (I2MTC), pp. 218-223, 6-9 May, 2013. [6] S. K. Kollimalla, M. K. Mishra, Adaptive Perturb & Observe MPPT Algorithm for Photovoltaic System, IEEE Power and Energy Conference at Illinois (PECI), pp. 42-47, 22-23 February, 2013. [7] H. El-hewl, M. Hassanien, H. A. Ashour, Maximum Power Point Tracking for Irregular Irridance of a Photovoltaic Array, 12 th
International Conference on Environment and Electrical Engineering (EEEIC), pp. 52-57, 5-8 May, 2013. [8] L. Cristaldi, M. Faifer, M. Rossi, S. Toscani, An Improved Model- Based Maximum Power Point Tracker for Photovoltaic Panels, IEEE Transactions on Instrumentation and Measurement, vol 63, no. 1, pp. 63- 72, January 2014. [9] S. Moballegh, J. Jiang, Modeling, Prediction, and Experimental Validations of Power Peaks of PV Arrays Under Partial Shading Conditions, IEEE Transactions on Sustainable Anergy, vol 5, no. 1, pp. 293-300, January 2014. [10] M. Balato, M. Vitelli, A Hybrid MPPT Technique Based on the Fast Estimate of the Maximum Power Voltages in PV Applications, 8 th
International Conference and Exhibition on Ecological Vehicles and Renewable Energies (EVER), pp. 1-7, 27-30 March, 2013. [11] B. Patnaik, P. Sharma, E. Trimurthulu, S. P. Duttagupta, V. Agarwal, Reconfiguration Strategy For Optimization of Solar Photovoltaic Array Under Non-uniform Illumination Conditions, 37 th IEEE Photovoltaic Specialists Conference (PVSC), pp. 1859-1864, 19-24 June, 2011. [12] M. Z. S. El-Dein, M. Kazerani, M. M. A. Salama, Optimal Photovoltaic Array Reconfiguration to Maximize Power Production under Partial Shading, 11 th International Conference on Environment and Electrical Engineering (EEEIC), pp. 255-260, 18-25 May, 2012. [13] Z. Cheng, Z. Pang, Y. Liu, P. Xue, An Adaptive Solar Photovoltaic Array Reconfiguration Method Based on Fuzzy Control, 8 th World Congress on Inteeligent Control and Automation, pp. 176-181, 7-9 July, 2010. [14] P. Dos Santos, E. M. Vicente, E. R. Ribeiro, Reconfiguration Methodology of Shaded Photovoltaic Panels to Maximize the Produced Energy, Brazilian Power Electronics Conference (COBEP), pp. 700- 706, 11-15 September, 2011. [15] G. V. Quesada, F. G. Gispert, R. P. Lopez, M. R. Lumbreas, A. C. Roca, Electrical PV Array Reconfiguration Strategy for Energy Extraction Improvement in Grid-Connected PV Systems, IEEE Transactions on Industrial Electronics, vol. 56, pp. 4319-4332, no. 11, November 2009. [16] M. Z. El-Dein, M. Kazerani, M. M. A. Salama, Optimal Photovoltaic Array Reconfiguration to Reduce Partial Shading Losses, IEEE Transactions on Sustainable Energy, vol. 4, pp. 145-154, no. 1, January 2013. [17] Y. Liu, Z. Pang, Z. Cheng, Research on an Adaptive Solar Photovoltaic Array Using Shading Degree Model-Based Reconfiguration Algorithm, Chinese Control and Decision Conference (CCDC), pp. 2356-2360, 26- 28 May, 2010. [18] D. Nyugen, B. Lehman, An Adaptive Solar Photovoltaic Array Using Model-Based Reconfiguration Algorithm, IEEE Transactions on Industrial Electronics, vol. 55, pp. 2644-2655, no. 7, July 2008. [19] K. S. Parlak, A New Reconfiguration Method for PV Array System, The 39 th Annual Conference of the IEEE Industrial Electronics Society (IECON 2013), pp. 1476-1481, 10-13 Nov, Vienna, Austria, 2013. [20] Y. Jiang, J. A. A. Qahouq, I. Batarseh, Improved Solar PV Cell Matlab Simulation Model and Comparison, Proceedings of 2010 IEEE International Symposium on Circuits and Systems (ISCAS), pp. 2770- 2773, May 30-June 2, 2010. [21] M. Seyedmahmoudian, S. Mekhilef, R. Rahmani, R. Yusof, E. T. Renani, Analytical Modeling of Partially Shadede Photovoltaic Systems, Energies, vol. 6, pp. 128-144, January 2013. [22] D. Picault, B. Raison, S. Bacha, J. Aguilera, J. De La Casa, Changing Photovoltaic Array Interconnections to Reduce Mismatch Losses: A Case Study, 9 th International Conference on Environment and Electrical Engineering (EEEIC), pp. 37-40, 16-19 May, 2010. [23] X. Wei, E. A. Spencer, Optimization of a Magnetosphere Model for Real-Time Space Weather Prediction Using a Modified Genetic Algorithm, IEEE Transactions on Plasma Science, vol. 38, pp. 2922- 2934, no. 10, October 2010. [24] M. Baygin, M. Karakose, Immunity-Based Optimal Estimation Approach for a New Real Time Group Elevator Dynamic Control Application for Energy and Time Saving, The Scientific World Journal, vol. 2013, Article ID 805343, 12 pages, doi: 10.1155/2013/805343, 2013.