Optimal Charging Strategy for EV Charging Stations by Two-Stage Approximate Dynamic Programming

This study targets the charging scenario for multiple chargeable vehicles at parking lots or charging stations where most vehicles have extended parking time, e.g. at the commercial buildings. All the vehicles are desired to be charged to full prior to the departure time specified, under constrained total charging capacity. A two-stage approximate dynamic programming (TSADP) framework is proposed to determine the optimal charging strategy, utilizing the predicted short-term future information and long-term estimation from historical data. The algorithm is separated into two stages, the optimization stage and approximation stage. The optimization stage is to find the optimal charging strategy for the short-term ahead time horizon. The approximation stage is to approximate the long-term time horizon charging cost, by the way of approximate dynamic programming (ADP). The whole algorithm works as a partial rollout algorithm and partial approximate dynamic programming. The TSADP framework can be integrated into stochastic programming framework, when the short-term information is uncertain with probabilities. The simulation results show that the proposed method can significantly decrease the energy cost.