Abstract: Aiming at the cold-chain logistics distribution vehicle routing problem under multi-distribution center, considering the timeliness and low-carbon cold-chain transport demand uncertainty, with the total objective of minimizing carbon cost, refrigeration costs, time penalty cost et al and maximizing the satisfaction of the customers, it sets up the fuzzy chance constrained model based on credibility measure theory. In the case of fuzzy demand, the improved genetic algorithm and large-scale neighborhood search is used to get the initial planned route, and then the actual demand is obtained by random simulation algorithm to adjust the routes. Finally, the effectiveness of the proposed algorithm and the rationality of the model are verified by comparison of other literature algorithms and example analysis, and the related parameters of the model are analyzed. Experimental results show that the improved genetic algorithm embedded with local search avoids local optimization and speeds up convergence. And when the demand is unknown, multi-distribution center cold chain distribution mode is better than single distribution center.

Key words: multi-depot vehicle routing problem, fuzzy demand, cold-chain, genetic-large domain search algorithm, carbon emission

摘要： 针对多配送中心下冷链配送的车辆路径问题，考虑到低碳冷链运输的时效性以及需求不确定性，以碳排放成本、制冷成本、时间惩罚成本等在内的总配送成本最小和客户满意度最大为总目标，建立基于可信性测度理论的模糊机会约束模型。在需求模糊的情况下，利用改进的遗传-大规模邻域搜索算法得到初始计划路线，再通过随机模拟算法得到实际需求，进行路线调整。通过与其他文献求解结果对比，以及分析算例来验证所建算法的有效性和模型的合理性，同时分析模型相关参数对目标值的影响。实验结果表明，嵌入局部搜索的改进遗传算法避免了局部最优，加快了收敛性。而且在需求不明时，多配送中心的冷链配送模式要优于单配送中心。

关键词: 多中心车辆路径问题, 模糊需求, 冷链, 遗传-大邻域搜索算法, 碳排放