An Efficient Approach for Appliances Energy Prediction using Differential Evolution Optimization and Light GBM Machine Learning Model

This study presents a new method for assessing the energy use of appliances in a low energy residence with the goal of improving energy consumption predictions. Information was collected at 10-minute intervals throughout a period of 137 days. The collected data includes a wide range of equipment, with a specific emphasis on lighting and the relationship between humidity levels and room occupancy. The dataset was divided into three subgroups depending on the lighting feature in order to assess the predictive importance of various parameters: all records, recordings without lighting, and records only including lighting. The suggested investigation employs a hybrid machine learning framework that integrates Extreme Gradient Boosting (XGBoost) and Light Gradient Boosting Machine (Light GBM) to forecast energy consumption. In order to improve the precision of the model, a sophisticated differential evolution approach is used to optimize the model's parameters. The models were compared using several performance indicators, such as the mean absolute error (MAE), root mean square error (RMSE), coefficient of determination (R2), and mean absolute percentage error (MAPE). The prediction results showed low error rates and strong correlation coefficients for all of these measurements, highlighting the usefulness of our technique in improving operational efficiency and forecast accuracy in energy management. The combination of bio-inspired optimization algorithms with ensemble machine learning approaches has significant promise for improving energy consumption predictions, which are essential for sustainable home energy management.