Medium and Short-Term Wind Power Prediction Based on ARIMA and LSTM Models

With the significant growth of green electricity trading globally, research on wind power generation has become particularly important. Wind power generation is characterized by its volatility and intermittency, which can affect the stability of the power system when connected in large scales. By achieving high-precision wind power forecasting, it is possible to optimize grid dispatching plans, reduce the phenomenon of wind curtailment, and ensure the safe and stable operation of the power grid. This study employs a hybrid method combining the Autoregressive Integrated Moving Average (ARIMA) model and the Long Short-term Memory (LSTM) model for wind power output prediction. Initially, the ARIMA model was used for preliminary forecasting, followed by further modeling of the ARIMA forecast results and original data using the LSTM model. Combining the linear modeling capability of ARIMA model with the LSTM model's ability to capture long-term dependencies enhances the forecasting performance, and the method was tested on wind power data from 50Hertz Energy Company in Germany.