ANN-based extraction approach of PV cell equivalent circuit parameters

In most practical cases, a PV array is constructed from a combination of standard PV units (solar cells or panels) connected in series and/or in parallel to obtain the desired power, voltage and current ratings. However, the non-linear behavior of solar cells or panels makes accurate determination of the model parameters a difficult task. A precise determination of the internal physical parameters of cells and panels is not always possible because of the non-uniqueness of the solution. As a result, some parameters of the PV panel model are estimated with non-negligible errors. In this paper, an advanced ANN-based determination approach of PV array parameters from individual cell/panel characteristics is developed for single-diode model. A proper selection of the training points for the ANN is also introduced to improve the estimation of the model. Simulation results obtained for simple case studies using Matlab/Simulink software are presented and discussed. The proposed ANN-based technique made possible more accurate determination of the parameters of the single-diode model with mean errors below 1% for all parameters except for the diode saturation and diffusion current which reached 11%.