Artificial Neural Network Based Modelling of Photovoltaic System

Photovoltaic system (PV) is alternate suitable option against conventional source of generating electricity. A typical photovoltaic system produces the electrical energy in terms of voltage and current. PV system is the environmental protection technique as well as saving of the electrical energy. This system reduces the level of chlorofluorocarbons (CFC) in the environment because it restricts the use of conventional method of generation of electrical energy. In this paper, an experimental study has been performed on PV system and modelling using artificial neural network. Artificial neural network (ANN) was constructed to predict symptoms of photovoltaic system for any climatic conditions. The symptoms of photovoltaic system were outlet voltage and outlet current. Feed-forward network was employed with LevenbergMarquardt back-propagation (trainlm) algorithm used as the training function. The outputs from the network were obtained and validated with the experimental results. Among the constructed networks, the best prediction performance was observed in two-hidden-layered network with minimum error. The modelling of the photovoltaic system was carried out with neural network toolbox of MATLAB with two inputs (Solar radiation and ambient temperature) and two outputs (Voltage and Current) value. The performance of the network was measured by mean square error (MSE). The proposed model may be used for any climatic conditions to predict the output from the photovoltaic system.