A local search with chain search path strategy for real-world many-objective vehicle routing problem

Abstract

This article focuses on a new application-oriented variant of vehicle routing problem. This problem comes from the daily distribution scenarios of a real-world logistics company. It is a large-scale (with customer sizes up to 2000), many-objective (with six objective functions) NP-hard problem with six constraints. Then, a local search with chain search path strategy (LS-CSP) is proposed to effectively solve the problem. It is a decomposition-based algorithm. First, the considered problem is decomposed into multiple single-objective subproblems. Then, local search is applied to solve these subproblems one by one. The advantage of the LS-CSP lies in a chain search path strategy, which is designed for determining the order of solving the subproblems. This strategy can help the algorithm find a high-quality solution set quickly. Finally, to assess the performance of the proposed LS-CSP, three instance sets containing 132 instances are provided, and four state-of-the-art decomposition-based approaches are adopted as the competitors. Experimental results show the effectiveness of the proposed algorithm for the considered problem.