Mathematical Modelling and Short-term Forecasting of Electricity Consumption of the Power System, with Due Account of Air Temperature and Natural Illumination, Based on Support Vector Machine and Particle Swarm

Abstract

The goal of this paper is to describe a forecasting model for the hourly electricity load. The model takes into account the meteorological factors (temperature and natural illumination) in the area covered by the Rostov utility dispatcher. In this study, support vector machine (SVM) with particle swarm optimization (PSO) were used to forecast electricity consumption. To get more accurate evaluation of the results of SVM model, the standard measures for quantitative evaluation of statistical performance and mean absolute percentage error (MAPE) were employed to evaluate the performance of various models developed. The results also suggest that the SVM method can be successfully applied to the forecasting model for the hourly electricity consumption in the area covered by the Rostov utility dispatcher.