SHORT-TERM LOAD FORECASTING BY USING THE ARTIFICIAL NEURAL NETWORK MODEL

Abstract

The article describes a model for short-term load forecasting by using the artificial neural network, and its application to load forecasting for a concrete distribution area. The artificial neural networks are mostly used in solving the problems of classification and prediction when the relations between input and output variables are highly complex and hard to describe exactly. Considering the stochastic nature and the major impact of weather conditions (temperature, humidity, wind, etc.) on electricity consumption, the application of artificial neural networks is suitable for short-term forecasting the load of an electric power system (EPS). Furthermore, the article describes the used models of artificial neural networks for seasonal and multiple daily load forecasts and presents the load forecast results for a distribution area supplied over the busbars of the 110/x substation at the HEP Elekroslavonija Distribution System Operator (HEP ODS).