Analysis and Prediction of Electric Energy Consumption Using a Deep Learning Approach: A Case Study of the Dessie District

Predicting electric power consumption is essential for modern energy management, addressing challenges like cost optimization, resource allocation, and sustainability. This study offers a thorough analysis of power consumption prediction to tackle the prevalent issue of inaccurate energy usage forecasts. A real dataset from the Ethiopian Electric Utility in the Dessie district, covering the years 2019 to 2023, forms the foundation of this research. Using advanced deep learning models, specifically Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and Bidirectional LSTM. This study proposes a robust methodology based on cutting-edge neural network architectures. The research includes detailed experimentation in data preprocessing, feature extraction, model development, and evaluation to showcase the potential of these models to transform energy management. The findings highlight these models' capabilities to improve operational efficiency, reduce costs, and enhance grid management. Despite challenges such as model overfitting and the need for precise hyperparameter tuning, model performance is evaluated using metrics like Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and Mean Absolute Error (MAE). Among the models, GRU demonstrated superior performance with minimal prediction error: 0.105 for MSE, 0.21 for RMSE, and 0.018 for MAE on testing data. This study emphasizes the potential of deep learning models to drive advancements in the energy sector, despite the existing challenges. Abyssinia Journal of Engineering and Computing, Vol. 4 No. 1 (2024)