Optimization of Time-Variant Charging Station Placement Using Evolutionary Algorithms

This paper introduces the optimization of time-variant placement of charging stations on a dynamic road network, with changing traffic conditions. We aggregated Uber Movement data from San Francisco into an hourly representation of a typical day, resulting in 24 snapshots of the network's traffic. The placement is optimized independently for each time interval by Evolutionary Algorithm, leading to distinct placements. that adapt to changing traffic conditions of each hour. Various network variables—length, speed, and travel time—are tested as heuristics, both as alternatives to optimization and for their effectiveness when combined with optimization. With enough charging stations, heuristic sampling yields near-optimal solutions, and combined with optimization, it slightly improves performance, with travel time outperforming the others.