Short Term Load Forecasting for Smart Distribution System Planning Using Deep Neural Networks: A Hybrid Approach

Abstract

Accurate load forecasting plays a crucial role in the management and control of electrical power in distribution systems. Short-Term Load Forecasting (STLF) is particularly vital for distribution planning, as it provides precise load predictions for the immediate future. This paper introduces an innovative hybrid deep-learning model specifically designed for STLF systems. The proposed hybrid model combines the strengths of Bidirectional Long Short-Term Memory (Bi-LSTM) and Gated Recurrent Unit (GRU) networks. The study utilizes a high-resolution real-world dataset, consisting of historical load consumption and weather-related features, with 30-minute intervals from the period of January 1, 2006, to December 31, 2010. This model is benchmarked against prominent standalone models such as Bi-LSTM, GRU, LSTM, and CNN, and hybrid models like CNN-LSTM and ConvLSTM-GRU. The model’s performance is evaluated using various validation metrics, including R-squared error, Root Mean Squared Error (RMSE), Mean Squared Error (MSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE). The results show that the proposed model outperforms all conventional models, offering significant improvements in forecast accuracy. Thus, the study highlights the potential of hybrid models in revolutionizing forecasting methodologies, paving the way for a smart distribution system.