Modeling of Output Power for Vehicle Charging System by improved K-means-LS-SVM

This research developed a charging system model based on an improved k-means-LS-SVM algorithm to deal with the influence of large-scale electric vehicle (EV) charging on grid dispatch. Firstly, we generated the output curve of the charging system by preprocessing the historical charging order data of the charging system. Second, we clustered the charging system's output curves using the improved K-means technique, resulting in a new data set of output curves that contained charging systems with comparable date clustering types for the three days before the test day and the test day. Finally, the photovoltaic output value of the test data was calculated by the LS-SVM algorithm. The results of the calculation example show that the improved k-means model combined with the LS-SVM algorithm can obtain better results in the output of the charging system. It is feasible to use the algorithm for establishing the charging system's output model.