The Method for Short-Term Forecast Electricity Load Based on Criteria Informativeness and Compactness

In the article research the method of preparing training samples for building the process of forecast electricity load. FRiS-Stolp function was used to clear the training samples from the noise component and emissions data. The optimal value of the parameter F* for eliminate noise is obtained. The results are presented in the charts. The forecast electricity load error for different values of the coefficient F* is also shown.