ABSTRACT
Photovoltaic parameter extraction has been the focus of many studies, each one presenting a new optimization method to obtain them. To further enhance the optimization process, this work presents a new cost function based on Total Least Squares and compares its performance with the usual function which is based on the Ordinary Least Squares approach. In this paper, eleven different metaheuristic methods are used to compare the performance of the two functions for both single and double diode photovoltaic cell models. The results are presented in terms of mean values of the estimated parameters. Additionally, the two cost function values are also evaluated according to their convergence properties for the different optimization methods considered. The results showed that the Total Least Squares method performs better parameter estimation than the Ordinary Least Squares approach, when used with all the eleven methods, where the best values are obtained when it is coupled with Teaching Learning Based Optimization algorithm for double diode model, resulting in a mean error value of 5.0375e-04 compared to 7.6423e-04 when using Ordinary Least Squares. The big difference between the convergence of the two cost functions was with the Dragonfly method. With the single diode model, this method results in a 0.0016 difference in the mean value of RMSE, and with the double diode model, this difference decreased to 0.0014.
Nomenclature and Abbreviations
Acknowledgments
The authors would like to thank BRO-CQ (ALT20-03-0247-FEDER-017659-BRO-CQ-IDT-COP-17659) for funding the work and ICT of University of Évora for enabling it. This work was also funded by FCT/MCTES through national funds and when applicable co-funded EU funds under the project UIDB/50008/2020.
Disclosure statement
No potential conflict of interest was reported by the author(s).