Abstract: There are problems of traditional genetic algorithm in solving multi-depot vehicle routing problem. First, variable chromosome length produced by conventional coding techniques leads to low computation efficiency and easily produces infeasible solutions. Second, parental genetic operators have less efficient during perturbation. And it is difficult to balance the relationship between elite proportion and population diversity, search depth and search breadth in different evolutionary populations. This paper designs a hybrid genetic algorithm to solve the problem, and the distribution network information is separately expressed in the encoding and decoding method to improve the computational efficiency. The control parameters of balanced elite ratio and population diversity are introduced in the selection operation. In addition, an adaptive search range strategy is proposed to effectively balance the relationship in both search depth and breadth. Through experimental results and comparative analysis, the proposed algorithm is verified. The research results provide a new method to solve the multi-depot vehicle routing problem and can also provide guidance for related logistics distribution decisions.

Key words: joint distribution; multi-depot vehicle routing problem; hybrid genetic algorithm; adaptive search range strategy