Forecasting of the electric vehicles' charging amount of electricity based on curves clustering

Abstract

The demand of charging in districts is one of the important influence factors to the siting and building of charging facilities for electric vehicles. Since the charging amount of electricity curves vary from district to district, a curves clustering method is proposed to simplify the forecasting process of differernt districts' charging amount of electricity. The proposed method takes the normalized charging amount of electricity curves as the input, then clustering those curves by hierarchical clustering method, and the clustering effectiveness is assessed by the sum of squares within cluster. Finally, this paper applies different predict methods for different clusters with peculiar features.