Real-time Locally Optimal Schedule for Electric Vehicle Load via Diversity-maximization NSGA-II

Abstract

As distributed energy storage equipments, electric vehicles (EVs) have great potential for applications in power systems. Meanwhile, reasonable optimization of the charging time of EVs can reduce the users' expense. Thus, the schedule of the EV load requires multi-objective optimization. A diversi-ty-maximization non-dominated sorting genetic algorithm (DM-NSGA) -II is developed to perform multi-objective optimization by considering the power load profile, the users' charging cost, and battery degradation. Furthermore, a real-time locally optimal schedule is adopted by utilizing a flexible time scale. The case study illustrates that the proposed DM-NSGA-II can prevent being trapped in a relatively limited region so as to diversify the optimal results and provide trade-off solutions to decision makers. The simulation analysis shows that the variable time scale can continuously involve the present EVs in the realtime optimization rather than rely on the forecasting data. The schedule of the EV load is more practical without the loss of accuracy.