A Modeling Method for Electric Vehicle Charging Load Based on Consensus K-means Clustering

Accurate prediction of electric vehicle (abbr. EV) charging load is the basis of optimization design for power supply planning. A clustering analysis-based EV charging load prediction method was proposed. The pinning consensus control was led in the clustering analysis of charging load data, and a consistency theory-based k-means clustering method was proposed. By use of dissimilarity measure of charging load data in current period and adjacent periods, the clustering state was iterated and updated and the clustering center was accurately calculated to complete the fast calculation of the probability distribution function of EV charging probability and starting charging time of EV. According to the identified characteristic parameters of EV charging behavior and by means of solving nonlinear planning function, the aggregated load model at charging peak time could be accurately predicted. Combining with practical case of Hangzhou EV charging, It is verified that the proposed method possesses the advantage of simple calculation, fast clustering and accurate modeling.