An Interaction-Enhanced Co-Evolutionary Algorithm for Electric Vehicle Routing Optimization

Abstract

Recently, the electric vehicle routing problem (EVRP) has emerged as a significant challenge in the transportation field. Unlike traditional vehicle routing problems, EVRP requires optimizing not only the customer route but also the charging scheme. This implies that routes must be planned while considering the availability of charging stations, and charging decisions must be made based on the specific route structure. Although the dual-population-based co-evolutionary algorithm (DPCA) has been proposed to address this coupling relationship, there is still room for further performance improvement. Therefore, this paper proposes an interaction-enhanced co-evolutionary algorithm (IECA) that facilitates a more effective collaborative search between routing and charging optimizations. For routing optimization, we propose an improved ant colony optimization method with a novel pheromone update approach to generate diverse route solutions. Specifically, we utilize information from the charging population to guide the pheromone update process. In contrast to DPCA, which solely relies on the best solution, we establish interaction among multiple solutions from both the routing and charging populations, which can better contribute the information interaction between two populations. Furthermore, we devise a novel population update strategy that enhances the selection of promising solutions. Experimental results validate the effectiveness of the proposed algorithm by demonstrating its ability to avoid local optima. Moreover, it achieves a reduction of approximately 4% in the average route distance across two public test suites.

Keywords: Electric vehicle routing problem, dual-population co-evolution, Ant colony optimization, enhanced interaction, pheromone update