Photovoltaic Power Prediction of BP Neural Network Based on Singular Spectrum Analysis

Aiming at the intermittent and fluctuating characteristics of photovoltaic power output, a BP neural network photovoltaic prediction model based on singular spectrum analysis is proposed. This method combines SSA and correlation analysis techniques, which can effectively improve the prediction accuracy of the model. First, SSA technology is used to decompose the photovoltaic output time series into trend series, oscillation series and noise series, and then the noise series with less influence are removed, and the trend series and oscillation series are predicted by the BP neural network model that takes into account similar days. Finally, the actual data in a certain area is taken as an example, and the prediction error analysis proves that the proposed method has higher prediction accuracy.