Medium to Long Term Wind Power Prediction Based on Improved mRMR Algorithm and GWO-LSTM Algorithm

With the advancement of China's electricity market reform, the development of new energy generation and power trading has entered a new stage. The new policy indicates that starting from April 2024, the grid will no longer bear the obligation of full-scale purchase, only purchasing a limited amount of guaranteed electricity, indicating that more electricity will enter the medium to long term market. The importance of the medium to long term electricity market is gradually highlighted. Wind power generation, as an important component of renewable energy generation, directly participates in medium to long term transactions in the electricity market. The increasing proportion of medium to long term electricity in wind power generation also poses higher requirements for medium to long term power prediction. Due to its obvious volatility, intermittency, and anti-peak characteristics of power generation, wind power prediction has become a challenging task in medium to long term power prediction. In this paper, based on the mRMR dimensionality reduction algorithm, medium term extended forecast meteorological data in the Shandong region are selected for feature extraction, and combined with the GWO-LSTM algorithm to achieve medium to long term wind power prediction, providing strong support for the declaration and scheduling control of medium to long term transaction curves in wind power.