Improved bat algorithm for Vehicle routing problem

Abstract

Vehicle routing problem (VRP) is the key issue of logistics system optimization. As a classical combinatorial optimization problem, it belonged to the typical NP-hard problem and remained unsolved. In this paper, the novel bat algorithm is proposed to solve VRP. The improvement is based on the combination of dynamic inertia weight and time factor. It can take full advantages of dynamic search by the random velocity and random step-size. Furthermore, with the real-number encoding approach, the discrete VRP can be converted into a quasi-continuous one. The procedure of the optimal searching in multidimensional continuous space can be implemented directly. Experimental results indicate that improved bat algorithm performs well for vehicle routing problem.