A Hybrid Ant Colony System for Partially Dynamic Vehicle Routing Problem

The constraints on responding time involved with dynamic vehicle routing problem (DVRP) require a compromise between distance and time in order to obtain the best solutions as fast as possible, so the flow time and the number of lost customers have gotten the mean to DVRP. Here, we propose a hybrid ant colony system (ACS) initially applied to design a static version of vehicle routing problem (VRP). Then, stochastic information of customers is used to be considered in the partially dynamic version of VRP. Our approach uses some heuristics to reconstruct routes and update pheromone. This approach for solving DVRP with the proper combination of parameters is compared with an insertion heuristic on the benchmark problems of Solomon's instances. The results show that our approach acts well in comparing some heuristic methods.