Artificial neural network based electric peak load forecasting

An artificial neural network (ANN) approach to electric peak load forecasting is presented. The ANN is used to learn the relationship among past, current, and future temperatures and peak loads. In order to provide the forecasted load, the ANN interpolates among the load and temperature data in a training data set. The average absolute error of 24-hour ahead forecasts in a test on actual utility data is shown to be 2.04%. Compared to other regression methods, the ANN allows more flexible relationships between temperature and load pattern.<
>