A practical model for single-step power prediction of grid-connected PV plant using artificial neural network

Abstract

A practical model for single-step power prediction of grid-connected photovoltaic plant is presented based on artificial neural network. The multi-dimensional running state space contains all relevant factors affect the power output is established after the pretreatment of the actual operating data of the photovoltaic plant. The predictive model using BP neural network is founded to predict the value of power with input variable include solar radiation, ambient temperature and other relevant factors. The model structure is fixed by cross validation. The hourly predictive value of power can be obtained from the neural network model step by step. The results of the 160kWp grid-connected photovoltaic plant in science and technology park of Yunnan Power Grid Corporation indicate that the proposed model performs well.