Photovoltaic power generation power prediction method, system, terminal and medium

Abstract

The application provides a photovoltaic power prediction method, system, terminal and medium, and solves the problem that although a traditional photovoltaic power prediction algorithm can capture short-term time sequence correlation, due to gradient disappearance, GRU and LSTM cannot capture longer-term time sequence characteristics in practice, and proposes an LSTC algorithm for photovoltaic power prediction, historical data is selected by a variable selection module, and after variable selection, short-term time sequence related data, long-term time sequence related data and non-time sequence related data are generated. Experimental results show that for photovoltaic power prediction, in a 24h prediction scenario, the MAE, RMSE and R2 of the LSTC are 0.8610%, 0.5288% and 0.7601% respectively, which are better than those of other algorithms. The application provides an effective solution for photovoltaic power prediction, which has important significance in the field of photovoltaic power prediction.

Images