A Comparative Study on Wind Power Forecasting Models Based on the Use of LSTM

Abstract

In the context of wind power generation's growing significance, this research tackles the critical problem of improving power system stability by reducing peak load and frequency control pressures using sophisticated wind power forecasting methods. Using the rapidly developing area of artificial intelligence and neural networks in particular, the study explores the efficacy of Long Short-Term Memory (LSTM), a recurrent neural network designed for event forecasting in time series data with long intervals and temporal delays. The research presents a novel forecasting model, LSTM that enhances prediction accuracy. This is corroborated by the calculated Root Mean Squared Error (RMSE) of 0.6782 and Mean Absolute Error (MAE) of 0.4614. The sustainability and dependability of wind energy integration into power systems may be enhanced by these findings, which highlight the possibility for more efficient and rapid convergence processes in wind power forecasting.