Rapid Calculation of Series and Shunt Resistance Values for a Solar Cell

Abstract

The series and shunt resistance values play an influential role in the output of a photovoltaic device. Additionally, they may also be useful parameters to assess the cell or module quality. Calculation of these values however is non-trivial, consequently a number of experimental and non-experimental techniques currently exist in order to estimate their values. In previous work [1], a method was proposed to calculate the values of series and shunt resistances at maximum power primarily using the experimental data typically published by the manufacturer. This numerical method based on the multi-dimensional Newton-Raphson method and current voltage equations expressed using the Lambert W-function is however computationally burdensome and therefore unsuitable for implementation in an environment where speed is needed such as manufacturing or field testing. Here we present a rapid method of calculation based on a feed forward type multi-layer perceptron artificial neural network (ANN). Current-voltage data were experimentally acquired for a single multi-crystalline silicon cell under varying levels of outdoor illumination from which the short circuit current, open circuit voltage, and location of the maximum power point values were obtained as typically published by the manufacturer, forming the training set for the ANN. The corresponding values of series and shunt resistances were then calculated using the method of Ghani and Duke [1] to form the network target set. It was found in this study that the network could be successfully trained for this application providing a method to rapidly calculate these loss values. It may therefore be a beneficial quality assessment tool for photovoltaic device manufacturers, installers, and users.