Capacitated Vehicle Routing Problems Based on Multi-heuristics with Balanced Search Control Particle Swarm Optimization

In this article, particle swarm optimization (PSO) is adopted to solve capacitated vehicle routing problems (CVRPs). In order to obtain high quality and diverse initial solutions to facilitate the subsequent search for optimal solutions by PSO, a sweep algorithm is combined with greedy randomized adaptive search procedure (GRASP) to produce high quality initial solutions, and an initial heuristic rate is also used to generate an initial solution with diversity. Meanwhile, a nonlinear S-decreasing curve is also employed to adjust the inertia weight of PSO to balance global and local searches. Moreover, local search and path relink with controlling rate are also involved to improve the quality of the solutions. Experiments were conducted on Augerat problems from the OR Library. Experimental results are compared with those from other studies and demonstrate that the algorithm designed in this study can effectively solve capacitated type vehicle routing problems.