We consider a general electric vehicle (EV) charging system with stochastic demand, demand request locations, and predetermined charging facilities (including charging station locations and charger capacities). The objective is to design a good routing strategy that accommodates well demand-request dynamics so as to satisfy the charging system’s stability constraints and also minimize the EV’s mean response time. We introduce a class of flexible and measurement-based routing policies called “partition-based random routing” (PBRR) and show that the performance measure of interest can be formulated as a constrained optimization problem with a convex objective function when the system is heavily loaded. This formulation enables us to establish strong theoretical results that are in aid of finding the optimal routing solution; however, in practice, finding this solution requires rather involved numerical calculations. To that end, we propose a surrogate, easy to design and implement, optimization algorithm for finding the desired optimal routing solution. Numerical work based on synthetic data shows that the performance of the developed routing strategy and its fast implementation is highly satisfactory for a number of system settings.

我们考虑具有随机需求、需求请求位置和预定充电设施（包括充电站位置和充电器容量）的通用电动汽车 (EV) 充电系统。目标是设计一个良好的路由策略，以适应良好的需求-请求动态，以满足充电系统的稳定性约束，并最大限度地减少 EV 的平均响应时间。我们引入了一类灵活且基于度量的路由策略，称为“基于分区的随机路由”（PBRR），并表明当系统重负载时，感兴趣的性能度量可以表述为具有凸目标函数的约束优化问题. 该公式使我们能够建立强有力的理论结果，以帮助找到最佳路由解决方案；然而，在实践中，找到这个解决方案需要相当复杂的数值计算。为此，我们提出了一种替代的、易于设计和实现的优化算法，用于找到所需的最佳路由解决方案。基于合成数据的数值工作表明，所开发的路由策略的性能及其快速实现对于许多系统设置都非常令人满意。