Search Results

Abstract: The increasing variation in residential electricity demand has created a need for accurate and efficient household electric power consumption forecasting for improved energy management and reliable operation of modern electrical distribution systems. Household electricity consumption is highly dynamic and nonlinear due to changes in appliance usage, consumer activities, seasonal conditions, and other influencing factors, making accurate forecasting challenging using conventional approaches. In this research, an efficient deep learning framework based on Bidirectional Long Short-Term Memory (BiLSTM) is developed for forecasting household electric power consumption. The Household Electric Power Consumption dataset is first preprocessed through missing-value handling, normalization, and feature selection to improve the quality of electrical load data. The processed data are then divided into training and testing sets, and the BiLSTM model is trained to learn temporal relationships within historical household electricity consumption patterns. The developed framework is implemented using Python Spyder 3.7 and evaluated using Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and R-Squared (R²) metrics. The experimental results achieve an MSE of 0.35, RMSE of 0.59, and R² of 0.47, demonstrating the capability of the BiLSTM model to capture important temporal characteristics of residential electrical demand. The developed framework can support household energy management, peak-load planning, demand-side management, and intelligent smart-grid applications.