Short-Term Residental DC Load Forecasting Using Extreme Gradient Boost (XgBoost) Algorithm
This research introduces a new approach for predicting short-term DC power loads using the XgBoost algorithm that utilizes weak learners to build robust prediction models using weaker learning combinations and shows the ability to capture complex non-linear patterns based on real-life residential DC load data.
Abstract
Accurate forecasts of short-term electricity consumption are essential for efficient energy management in buildings and residential households. This research introduces a new approach for predicting short-term DC power loads using the XgBoost algorithm. This method includes feature selection based on tendencies and patterns in historical data. The XgBoost algorithm utilizes weak learners to build robust prediction models using weaker learning combinations and shows the ability to capture complex non-linear patterns based on real-life residential DC load data. The simulation indicates that XgBoost algorithms exceed conventional forecasting methods' results in precision when trained after the data analysis-driven features selection. Evaluation metrics such as the root mean square error (RMSE), the mean absolute percentage error (MAPE), the mean square error (MSE), and the mean absolute error (MAE) are used to measure forecast accuracy. The proposed approach showed considerable improvements in the delivery of 24-hour DC electricity load forecasts to residential consumers.