Wind power prediction based on long short-term memory neural network

Electricity produced by the wind is a significant way to utilize renewable wind energy. Because of the volatility and uncertainty of wind power, existing forecasting systems need to significantly increase their prediction accuracy. Therefore, this paper suggests a Long short-term memory neural network-based wind power forecast model (LSTM). First, data are gathered to identify the model's input and output; next, an LSTM-based wind power prediction model is built; this creates a mapping between the model's input and the output of the predicted wind power; and finally, a case simulation analysis is applied to predict the wind power in accordance with the results. Compared to the GRU, ANN, and BP models, the LSTM-based wind power prediction model established in this paper's MAE evaluation index is reduced by 7.18%,31.74% and 17.83%. So the model established in this paper can improve the reliability and effectiveness of wind power prediction.