Separable Nonlinear Least Squares Search of Parameter Values in Photovoltaic Models
Jie Xu
Abstract
This article presents a new numerical approach for parameter extraction in five-parameter single-diode or double-diode photovoltaic models, through a separable nonlinear least squares treatment of the problem. The parameter separation technique realizes a reduction of the number of independent parameters as well as the computational complexity. Then, a penalty-assisted direct search algorithm is proposed to solve the optimization problem. The proposed method is applicable to current <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">–</i> voltage ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">I–V</i> ) curves with at least 4 points or manufacturer datasheets with 3 key points, which breaks the 5-point limitation in previous numerical optimization methods. In two commonly-used case studies that are considered as the benchmark test for single-diode model, the proposed method shows a high accuracy close to the most accurate methods. After an additional curve-fitting refinement of the obtained parameters, the best accuracy documented until now is achieved in the case studies. Furthermore, the robustness of the proposed method is validated through a large-scale data repository containing more than one million <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">I–V</i> curves. Finally, to facilitate the application of the proposed method, a graphical easy-to-use program has been developed and available to public users.