Forecasting Power Output of Photovoltaic System Using A BP Network Method

Abstract

The characteristics of intermittent and stochastic of solar energy has brought great challenges to power grid system in terms of operation and regulation. Power forecasting is an important factor for optimal schedule of power grid system and assessing the working performance of PV systems. In order to forecast the power output of a PV system located in Ashland at 24-hour-ahead for higher efficiency, a back propagation (BP) neural network model is proposed. Before designing the model, correlation analysis is done to investigate the relationship between power output and solar irradiance and ambient temperature, which are key parameters affecting the power output of PV systems. Based on a correlation analysis, the model admitted the following input parameters: hourly solar radiation intensity, the highest, the lowest daily and the average daily temperature, and hourly power output of the PV system. The output of the model is the forecasted PV power output 24 hours ahead. Based on the datasets, the neural network is trained to improve its accuracy. The best performance is obtained with the BP neural network structure of 28-20-11. The analysis of the error indicator MAPE shows that the proposed model has great accuracy and efficiency for forecasting the power output of photovoltaic systems.