Short-Term Load Forecast using Wavelet Transformation

摘要

In this paper an approach how wavelet transformation can be used for load forecast is presented. The load data is transformed in low and high frequency components. It is shown that the high frequency component does not change from a reference day to the forecast day. Whereas the low frequency component is situation (climate) dependant. Hence a forecast model is developed for the low frequency component only. A time series approach is used. The parameters of the model are computed by the recursive least squares method. As an example application, the short-term load forecast for a public utility is described. Compared to the up to now used methods the accuracy is enhanced for all horizons.