Min-max Vehicle Routing Problem Based on Ant Colony Algorithm

To minimize the length of travelling distance of the longest sub-route in vehicle routing problem, the max-min ant system with parameter adaptation is adopted, which can be applied to different datasets in practice. Routes are constructed by sequential and parallel methods for the customers with clustering and random distribution respectively. Since the behavior of ant colony algorithm depends strongly on the given parameter values, these parameters include expectation heuristic factor, choice probability, level of pheromone persistence, and the number of ants, are self-adaptive at different stages in the course of algorithm execution, which help to accelerate convergence and enhance the searching around optimal solution, as well as to guarantee the diversity of solution to avoid falling into local optimization. Seven classical instances are tested for min-max vehicle routing problem; the results demonstrate that max-min ant system with parameter adaptation has high effectiveness, fast convergence speed, and robustness in solving these problems.