Short term load forecasting using artificial intelligence

Abstract

This paper presents a comparative study of short-term load forecasting using Artificial Intelligence (AI) and the conventional approach. A feed-forward, multilayer artificial neural network (ANN) was employed to provide a 24-hour load demand forecast. In this model, historical data, weather information, day types and special calendar days were considered. The forecasted results using AI were compared with those of conventional method. From the simulations it is found that the maximum forecasting percentage error for AI is approximately 5.5% as opposed to 15.96% for the conventional approach.