SSA-LSTM: Short-Term Photovoltaic Power Prediction Based on Feature Matching

To reduce the impact of volatility on photovoltaic (PV) power generation forecasting and achieve improved accuracy, this article provides an in-depth analysis characteristics PV outputs under typical weather conditions. The trend similarity between simultaneous are found, a hybrid prediction model based feature matching, singular spectrum (SSA) long short-term memory (LSTM) network is proposed. In paper, correlation used to verify generation; days historical meteorological data calculated through grey relation analysis; similar generated levels searched for phase matching. input time series decomposed by component, oscillation component noise extracted; principal reconstruction carried out each component. Then, LSTM established reconstructed subsequences, external controlled compare obtained results. Finally, performance evaluated plant in certain area. experimental results prove that SSA-LSTM has best performance.