Enhanced Wind Speed Prediction Using Dual-Memory LSTM: A Novel Approach to Temporal Dynamics

Abstract

Accurate wind speed forecasting is crucial for optimizing wind energy integration, ensuring grid stability, and advancing renewable energy systems. This study introduces the Dual-Memory Long Short-Term Memory (DMLSTM) model, an innovative extension of conventional LSTM architectures designed to model short-term and long-term temporal dependencies explicitly. By incorporating separate memory cells and dynamic gating mechanisms, the DMLSTM addresses the limitations of traditional models such as AutoRegressive Integrated Moving Average (ARIMA), Artificial Neural Networks (ANN), and baseline LSTM in capturing the non-linear, stochastic, and hierarchical patterns inherent in wind speed data. The research utilized a comprehensive meteorological dataset from Tetouan City, comprising wind speed, temperature and humidity. Preprocessing steps ensured data quality and consistency, including data normalization and outlier handling. The DMLSTM was trained using a sliding window approach, mapping sequences of historical data to predict future wind speeds, and evaluated against ANN, ARIMA, and baseline LSTM models using Root Mean Square Error (RMSE) and Mean Absolute Error (MAE). Experimental results demonstrate the superior performance of the DMLSTM, achieving the lowest RMSE and MAE values across all tested models. Visual comparisons of predicted and actual wind speeds reveal the DMLSTM's robustness in capturing abrupt changes and complex temporal dynamics. The findings underscore the model’s potential as a reliable tool for wind speed forecasting, paving the way for more effective integration of wind energy into sustainable energy systems. This study highlights the significance of advanced temporal modeling techniques in addressing the challenges of renewable energy forecasting.