Short-term load forecasting for demand side management

Abstract

A method for short-term load forecasting which would help demand side management is presented. This is particularly suitable for developing countries where the total load is not large, especially at substation levels, and the data available are grossly inadequate. It is based on the Kalman filtering algorithm with the incorporation of a ‘fading memory’. A two-stage forecast is carried out, where the mean is first predicted and a correction is then incorporated in real time using an error feedback from the previous hours. This method has been used to predict the local load at 11 kV and also the bulk load at 220 kV. The results and the prediction errors are presented.