Prediction of Solar Irradiance Based on BP Neural Network with Wavelet Decomposition

To improve the precision of solar irradiation (SI) prediction, this paper presents an SI prediction method based on BP neural network (BPNN) with wavelet decomposition (WD). Firstly, analyze data by wavelet, and WD applied to train input set, training target set, and testing target set. Then combined with BPNN to train, predict, and normalize the decomposed data. Finally, the anticipated results of the two methods are output as predicted values. The prediction results show that the proposed manner is consistent with the expected value curve when predicting SI, which improves the prediction accuracy and provides a basis for SI prediction of photovoltaic power plants.