The accelerated electric vehicle (EV) adoption rate has worsened the congestion in the electric charging stations, which results in the elongation of wait times, poor utilization of charging station, and unneeded travel detours. As such, it is paramount to optimize schedules of charging stations to enhance the user experience and the performance of the stations. The present work deals with assigning EV charging stations to a road system, the task of which is to reduce the average waiting time, travel distance, and loss of parts of energy in the conditions of the station capacity. To find a solution to the scheduling problem, a population-based metaheuristic, denoted exponential convex optimization algorithm (ECOA) is proposed. To smooth short-term variations, which made the solution update mechanism more unstable, ECOA also adds an exponentially weighted moving average (EWMA) to the convex optimization algorithm (COA). EWMA is used as a smoothing trajectory-level technique as opposed to a learning or prediction model. In contrast to the approaches that are entailed by reinforcement learning, the proposed method does not include any training information, reward modeling, or policy training and can be deployed in real-time. The findings of simulation using different traffic loads make it very clear that ECOA is always better than the baseline scheduling heuristics in minimizing the waiting time and travel distance without compromising on the feasibility of the energy limits. Such findings suggest that ECOA is an effective and computationally efficient problem solver of EV charging station scheduling in dynamic settings.
Bhuvaneswari et al. (Sat,) studied this question.