Artificial Neural Network Training of Unshaded Datasheet for Photovoltaic Maximal Power Point Prediction under Partial Shading Conditions

Partial shading conditions (PSCs) significantly reduce photovoltaic generated power and increases the difficulty of maximal power point (MPP) tracking. In this paper, artificial neural network (ANN) model is adopted to predict the photovoltaic MPP, so as to address the issue of MPP tracking under PSCs. Firstly, the irradiance, temperature, and maximum output power of photovoltaic system under uniform irradiation are collected, and the constructed ANN is analyzed and trained. After that, this ANN trained by unshaded datasheet is used to predict the MPP under PSCs. To increase accuracy in this step, the prediction results are arranged in ascending order of error, and the MPP under PSCs is predicted again. This error ranging process greatly reduces the prediction error. According to the simulation results, changing the number and sequence of input data of ANN greatly affects the error of prediction results. Therefore, this paper realizes accurate prediction of photovoltaic MPP under PSCs by collecting the datasheet of photovoltaic system without shading. The proposed method is beneficial to reducing the ANN training cost and enhancing its prediction accuracy in scenarios drastically different from the training context.