Power system yearly peak load forecasting: a grey system modeling approach

Abstract

Applications of grey system modeling techniques to predict power system yearly peak loads and their occurring dates are described in this paper. For yearly peak load forecasting, a new hybrid grey model of GM(1,1) in conjunction with GM(1,2) is proposed. Corresponding dates on which the peak loads occur are predicted by using the topological forecasting method. Complying with the characteristics of minimal historical yearly peak load records (only one peak load value for a year), the grey system techniques used in this paper show that a small amount of historical data (3 to 7 points of data) are required to set up a high forecasting-accuracy model. Effectiveness of the developed models has been demonstrated through predicting the actual Taiwan Power (Taipower) yearly peak loads and the occurring dates.