An Improved Artificial Bee Colony Algorithm For The Two-echelon Vehicle Routing Problem

In this paper, we propose an artificial bee colony algorithm (ABC) based on largeneighborhood search for the capacitated vehicle routing problem (CVRP) and two-echelonvehicle routing problem(2E-VRP). CVRP is a basic and classic model of vehicle routingproblem based on single layer delivery, while the2E-VRP problem arises in two-leveltransportation systems that encountered in the context of city logistics. Unlike the CVRP inwhich freight is directly transported to customers, freight arrives at a major terminal and istransported through intermediate satellite platforms to the final customers. Largeneighborhood search (LNS) method has recently been proved to be outstanding in solvingtransportation and scheduling problems. ABC algorithm is a new bionic swarm intelligencealgorithm and is known for its strong global optimization. So we use the LNS asneighborhood operators to improve ABC algorithm.CVRP is the base of2E-VRP, so we proposed an ABC algorithm based on largeneighborhood search to solve the CVRP. Experiments showed that our improved ABCalgorithm is effective to CVRP and is more stable than the original ABC algorithm. We alsoproposed an ABC algorithm based on large neighborhood search for the2E-VRP, which is themajor study in this paper. The algorithm searches vehicle routes of each echelon respectivelyto find a better neighbor by destroying and repairing the current solution iteratively.Experimental results show that our method performs much better than other known methodsfor2E-VRP.