A load forecasting method for HEMS applications

Abstract

In a home energy management system (HEMS), household load forecasting is difficult, due to its small number of loads and random nature of turning on/off. However, it is important to pre-schedule the load demands of home appliances in the HEMS for power expenditure minimization. This paper proposes a new day-ahead short-term artificial neural network (ANN) based forecasting method, which consists of the techniques of data selection, wavelet transform (WT), ANN-based forecasting, and error-correcting (EC) functions. To verify the effectiveness of the proposed forecasting method, the approach has been verified by using practical data for household load demands. Numerical forecasting results are presented and discussed in this paper.