To satisfy no delay and minimize the travelling distance of all routes for dynamic vehicle routing problem with time window, the planning horizon is divided and the dynamic vehicle routing problem is partitioned into a series of static subproblems, which are solved by improved max-min ant system. In the max-min ant system, routes are constructed with the parallel and sequential methods for datasets respectively aiming at customers in random and clustering distribution. As the behavior of max-min ant system depends largely on the value of parameters, multiparameters including distance heuristic factor, choice probability, level of pheromone persistence, level of time window and the number of ants are self-adapted at different stages in the course of algorithm execution. Instances are tested and results show the effectiveness in solving these problems.
展开▼