Electric Vehicle Charging Facility Planning Method Based on Curves Clustering Forecast

In view of the planning and construction of electric vehicle (EV) public charging facilities, a charging demand forecast model is established with the objective of the size of charging facilities meeting the charging demand of EV. The cluster prediction method is applied based on clustering of charging curves, the clustering trend of charging curves is evaluated by Hopkins statistic, and cluster numbers are selected reasonably by elbow method. Then, the trend and irregular characteristics of every clusters are investigated and appropriate predict method, such as Holt-Winters exponential smoothing and autoregressive integrated moving average (ARIMA), is chosen. In the end, charging curves of EV from 15 counties and districts of one city are taken as simulation case, the analysis results show that the proposed model increased the predict accuracy of charging demand effectively, and increased the precision of siting and sizing of public charging facilities.