Random Scenario-Based Dynamic Location Optimization for EV Fast-Charging Station

The growing popularity of electric vehicles (EVs) warrants the optimization of fast-charging station locations for mitigating station location range anxiety. In this study, a location model for EV fast-charging stations (EVFCSs) is proposed to minimize the total travel time, and a path planning method based on reinforcement learning (RL) is presented. Random scenarios were generated in this study to simulate the real charging behavior of EVs considering the uncertainty of driver behavior and other factors, such as travel, path planning, and EVFCS selection. The method reflects the randomness of the charging scenario by generating multiple alternative paths and changing the target charging station and driving route when the vehicle travels to other nodes. The point demand and flow demand models are combined to establish an EVFCS optimal location method by improving the structure of a multipopulation genetic algorithm (MPGA). Based on the gradual increase in the penetration rate of EVs, a multistage location planning for EVFCS was developed. According to tests using random scenarios, the obtained location planning showed good robustness; thus, it optimizes the driver travel costs and benefits the EVFCS operator.