Open Access
ARTICLE
Hybrid Machine Learning Models for Energy Consumption Forecasting in Intelligent Power Grids
Issue Vol. 1 No. 01 (2025): Volume 01 , Issue 01 Oct 2025 --- Section Articles --- Published Date: 2025-10-01
Abstract
Accurate energy consumption forecasting has become essential for maintaining stability, efficiency, and resilience in modern intelligent power grids. Traditional statistical forecasting models often struggle to handle nonlinear consumption patterns caused by climate variability, distributed renewable energy, consumer behavior, and fluctuating demand curves. This research investigates the use of hybrid machine learning models, combining classical predictors (ARIMA, SARIMA) with advanced learning algorithms (LSTM, GRU, XGBoost, Random Forest), to enhance forecasting accuracy and grid responsiveness. Using high-resolution time-series datasets, the study demonstrates that hybrid approaches significantly outperform standalone machine learning models by leveraging both linear and nonlinear patterns. Results reveal improvements in prediction accuracy (RMSE, MAE), peak-load anticipation, anomaly detection, and demand-side optimization. The findings support hybrid ML systems as a powerful solution for next-generation smart grids.
Article Cover Image
