Research On The Short-term Prediction Of Photovoltaic Power Output

Since the energy shortages and environmental problems, photovoltaic power has been developed rapidly. Because the power output of PV system is random, the growing of PV system integrated to power grid has greatly influence on the security and stability of power grid.Therefore, scientifically and accurately predict the power output of PV system is of great significance for promoting the security and stability of the power system.This paper study on the short-term prediction of photovoltaic power output according to the problems.Based on analysis of the PV system output characteristics and the main factors affecting the PV output, BP neural network forecasting model were established according to the type of season. The experiment results show that: the prediction accuracy of classification forecasting model is significantly higher than traditional unified forecasting model. BP neural network depends on the number of training samples and sample distribution, for it based on empirical risk minimization. Based on the analysis of the similarity date theory,this paper established support vector machine(SVM) forecasting model. The experiment results show that: the SVM forecasting model can accurately predict the PV power output in small samples.Finally, a forecasting system was designed and realized,which has the basis function of PV forecasting and is practical.