An Efficient Monthly Load Forecasting Model Using Gaussian Process Regression

Abstract

Load forecasting is a powerful tool which helps the electric utility to make important decisions such as purchasing and generation of electric power, scheduling, load shedding/switching, and infrastructure development. Although a wide range of mathematical and computational approaches is available for load forecasting, still there is a scope of improvement in the methodologies implemented over real-time data. In this paper, a load forecasting model based on Gaussian Process Regression (GPR) method is presented. The GPR model is a nonparametric model with a kernel function and is trained with a set of large datasets containing hourly loads of an Australian city. The performance of the proposed model is assessed in terms of the Mean Average Percentage Error (MAPE). Moreover, the obtained results are compared with the results obtained from Decision Tree (DT) and Artificial Neural Network (ANN) models. The comparative study shows the superiority of the GPR model.