A Novel deep learning model to improve electric energy consumption forecasting accuracy for lighting, residential, and commercial loads in the mediterranean region

Abstract

Accurate electricity load forecasting (ELF) is essential for ensuring efficient energy management and grid stability. This study aims to improve forecasting accuracy for sectoral electricity consumption in the Mediterranean region, specifically in Adana, Mersin, and Antalya. To achieve this, deep learning (DL) models including long short-term memory (LSTM), convolutional neural network (CNN), and a hybrid CNN-LSTM model are developed using monthly data from 2016 to 2023. A grid search approach is employed to optimize key hyperparameters such as batch size, lookback window, dropout rate, and network structure. The performance of the models is evaluated using root-mean-squared error (RMSE), mean absolute error (MAE), mean absolute percentage error (MAPE), and coefficient of determination (R2) metrics. The results indicate that the hybrid CNN-LSTM model outperforms standalone CNN and LSTM models across all load types, achieving higher accuracy and better generalization. Furthermore, a MATLAB/Simulink-based dynamic power system model is developed to analyze the interaction between power generation and load demand. The findings demonstrate that the proposed approach provides a reliable and effective framework for sector-based ELF and supports improved decision-making in energy management and planning.