Ultra short-term forecast of photovoltaic generation based on time-division LSTM neural networks

Abstract

Accurate forecast of photovoltaic generation is important for power system operation. This paper proposes a new ultra short-term forecast of PV generation based on Spearman correlation coefficient analysis and time-division long short-term memory network. The method first uses the Spearman correlation coefficient to analyze the factors affecting photovoltaic power at each moment. The influencing factors with high correlation are selected as the input variables of the long short-term memory network model. A forecasting model based on long short-term memory network is established for each moment to realize the forecasting of time-division photovoltaic generation. Finally, case study is performed using historical generation output of a practical PV plant. Study results indicate that the proposed method has higher forecast precision than the single long short-term memory network forecast model.