Daily load curve forecasting via automatic selection of models

A forecasting method which can be easily implemented on a computer is described. The development involves an objective way of representing the relationship between the load and its determining factors. In the paper a forecasting method is proposed which uses automatically selected models and objective descriptions for accounting for the effects of temperatures and day kinds e.g. weekdays and weekends, on daily load curves. A daily load curve forecasting method is proposed. In the method, the best load model is selected automatically based on a criterion, and the special days effect on the load is classified and modeled. Thus we can treat the time varying nature of the load and special days effects in a suitable and effective way.