Extraction of the Multijunction Solar Cell Parameters Using Two Metaheuristic Algorithms

The module with multijunction solar cells (MJSC) is an excellent solution for converting solar radiation into electrical energy. Several methods are applied to extract the parameters of the multijunction solar cell. Most of them are analytical and numerical. Metaheuristic algorithms have lately been used for the parameters' extraction. In specialized literature, only a few multijunction solar cells have been performed to extract the parameters. Therefore, two metaheuristic algorithms are proposed in this paper: the Chameleon Swarm Algorithm (CSA) and the Black Widow Optimization Algorithm (BWOA). The first algorithm (CSA) is applied for the first time to estimate the solar cell and panel parameters, and both algorithms (CSA & BWOA) are employed for the first time for the multijunction solar cell. They are applied considering two models: the single diode model (SDM) and the double diode model (DDM) for the multijunction solar cell, and three work temperatures of the multijunction solar cell. Four statistical tests are used to analyze the performance of the algorithms, the main being root mean square error (RMSE). A comparative study was performed using the other analytical and metaheuristic algorithm. The obtained RMSE is 8.9120260701E-5 for BWOA and 8.9123932518E-5 for CSA model, respectively, in the case of SDM model and 41.5°C, 1.0186359636E-4 for BWOA and 1.0088314434E-4 for CSA model, respectively, in the case of DDM model and 41.5°C. In the case of the solar panel the RMSE is 3.62E-3 for BWOA, 3.6250156794E-3 for CSA model, respectively, in the case of SDM model and 25°C. The best root mean square error results are obtained using the Black Widow Optimization Algorithm for the single-diode model. The lowest value for root mean square error is 8.9120260701E-5. The special feature and merits of the proposed algorithms are that they have better exploration and exploitation ability; thus, they provide the optimal results with reduced computational time. Further, the performance of the two algorithms (BWOA andCSA) is validated using the dataset of the CTJ 30 panel. The BWOA algorithm has a root mean square error that is two times lower than the one in the research literature. The computational time is also calculated. It is around 2 s, which is very competitive for all considered cases. CSA has the lowest computing time for all four cases considered, varying from 1.882829 s to 2.277469 s. Furthermore, variation in the temperature function is studied using the extracted parameters.