Composite modeling for adaptive short-term load forecasting

A composite load model is developed for predicting hourly electric loads 1-24 h ahead. The load model is composed of three components: the nominal load, the type load, and the residual load. The nominal load is modeled in such a way that the Kalman filter can be used, and the parameters of the model are adapted by the exponentially weighted recursive-least-squares method. The type load component is extracted for weekend load prediction and updated by an exponential smoothing method. The residual load is predicted by the autoregressive model, and the parameters of the model are estimated using the recursive-least-squares method. Test results are presented using utility data for two different years.>