Relocation Matching for Multiple Teams in Electric Vehicle Sharing Systems

Abstract

This paper designs a relocation scheduler for electric vehicle sharing systems, aiming at overcoming stock imbalance and enhancing service ratio by evenly distributing relocation load for multiple service teams. To exploit genetic algorithms, a feasible schedule is encoded to an integer-valued vector having k+m-1 elements, where k is the number of vehicles to move and m is the number of service teams. Two indices are built for overflow and underflow stations, making each vector element denote a source and a destination by its position and the value itself. In addition, negative numbers are inserted to separate the subschedules for each team. The maximum of relocation distances is calculated in the cost function while the genetic iterations reduce the cost generation by generation. The performance measurement result, obtained by a prototype implementation, finds out that each addition of a service team reduces the relocation distance to 47.3 %, 32.0 %, and 25.0 %, making it possible to tune the system performance according to the permissible budget and available human resources.