Medium and Long-Term Power Generation Forecasting and Operation Strategies for Wind Farms Based on Deep Learning

The power generation of wind farms is highly volatile and stochastic due to the influence of meteorological factors, and there is a lack of scientific operation and maintenance strategies. Based on the meteorological factors affecting the power generation of wind farms, this paper employs the LSTM deep learning method of multiple time series to forecast the medium and long-term power generation of wind farms and proposes a predictive operation and maintenance mode that combines power generation forecasting and the equipment status assessment system. Among them, the principal component analysis method is adopted to extract key influencing factors, and the collected data of a certain wind farm is integrated into medium and long-term data to obtain the forecasted power generation. It has been verified that the deep learning forecasting evaluation indicators such as RMSE are in line with expectations, the forecasting results are favorable, and a predictive operation and maintenance system architecture is proposed in combination with deep learning to optimize the quarterly power generation and guide precise operation and maintenance.