ENERGY CONSUMPTION FORECASTING MODELS FOR SMART GRIDS: A STATE-OF-THE-ART REVIEW AND APPLICATION PERSPECTIVES

Energy consumption forecasting plays a crucial role in optimizing the operation of modern power systems, particularly in the context of smart grids and renewable energy integration. Accurate prediction models enable utilities and policymakers to plan generation, balance demand, and improve energy efficiency. This paper presents a comprehensive review of forecasting approaches ranging from traditional statistical methods (ARIMA, SARIMA) to machine learning algorithms (SVM, Random Forest, KNN) and deep learning architectures (ANN, MLP, LSTM). The main contribution of this review is to highlight the strengths, limitations, and application contexts of these models according to the nature of available data, the time horizon, and computational constraints. A comparative synthesis of recent studies (2020 2024) is provided, focusing on evaluation metrics such as RMSE, MAE, MAPE, and R. Finally, perspectives are proposed for hybrid and data-driven approaches, particularly for developing countries where data scarcity and climatic variability remain major challenges.