Optimizing electricity demand forecasting with a novel RNN-LSTM hybrid model

Abstract

Accurate electricity demand forecasting is essential for ensuring the reliability, efficiency, and sustainability of modern power systems. In this study, a hybrid deep learning model combining Recurrent Neural Networks (RNN) and Long Short-Term Memory (LSTM) units is proposed to predict short-term electricity demand with high precision. The model utilizes historical consumption and generation data to capture both short-term fluctuations and long-term temporal dependencies in electricity usage patterns. Comprehensive experiments were conducted, and the model’s performance was evaluated using standard metrics such as R2, Mean Absolute Error (MAE), Mean Squared Error (MSE), and Root Mean Squared Error (RMSE). The proposed RNN-LSTM model achieved remarkable results, with an R2 score of 0.9901, MAE of 0.0123, and RMSE of 0.0187, outperforming standalone RNN, LSTM, and several benchmark models in the literature. The results demonstrate the model’s ability to accurately learn complex demand dynamics and make reliable forecasts, which are critical for demand-side management and optimal resource allocation in smart grid systems. This study provides valuable insights into hybrid deep learning approaches for energy forecasting and establishes a foundation for future research that could integrate multivariate features, external conditions, and real-time predictive control mechanisms for enhanced performance in practical applications.

Topics & Concepts