An Improved Artificial Neural Network Method for Predicting PV Output Performance

In order to ensure the safety and stability of the power grid with photovoltaic (PV) generation integration, it is necessary to predict the output performance of PV module under varying operating condition. In this paper, an improved artificial neural network (ANN) method is proposed to predict electrical characteristics of PV module by combining several neural networks under different environmental conditions. In order to study the dependence of output performance on solar irradiance and temperature, the proposed neural networks model is composed of four neural network. Each neural network consists of three layers, in which input is solar radiation and module temperature and output are five physical parameters of single diode model. The experimental data is divided into four groups and used for training the neural networks. The electrical property of PV modules, including I-V curves, P-V curves and normal root mean square error, are obtained and discussed. The effectiveness and accuracy of this method is verified by the experimental data for different type photovoltaic modules. Comparing with traditional single ANN method, the proposed method shows better accuracy under different operating conditions.