A Short-Term Wind Speed Forecasting Model Based on a Multi-Variable Long Short-Term Memory Network

Accurately forecasting wind speed on a short-term scale has become essential in the field of power energy. In this paper, multi-variable long memory network model (MV-LSTM) based Pearson correlation coefficient feature selection is proposed to predict speed. The method utilizes multiple historical meteorological variables, such as speed, temperature, humidity, and air pressure, next hour. Hourly data collected from two ground observation stations Yanqing Zhaitang Beijing were divided into training test sets. sets used train model, evaluate with root-mean-square error (RMSE), mean absolute (MAE), bias (MBE), percentage (MAPE) metrics. compared other methods (the autoregressive moving average (ARMA) single-variable (LSTM) method, which inputs only data) same dataset. experimental results prove feasibility MV-LSTM for its superiority ARMA LSTM method.