基于改进灰狼算法的生鲜农产品多车型配送路径优化研究

Research on Multi-Vehicle Delivery Route Optimization of Fresh Agricultural Products Based on Improved Gray Wolf Algorithm

摘要: 在生鲜农产品消费需求持续增长和冷链物流精细化运营要求不断提高的背景下，企业配送活动面临配送成本高、车辆资源匹配不足、路径安排不合理和客户服务水平难以兼顾等问题。针对生鲜农产品配送中多车型协同、重量与体积双重约束以及时间窗服务要求并存的实际特点，本文构建了考虑运输成本、车辆固定成本、制冷成本、货损成本、时间惩罚成本和客户满意度的多车型车辆路径优化模型，并设计改进灰狼算法进行求解。算法采用随机键编码方式表示客户访问顺序和车辆启用顺序，通过约束解码实现客户分配与车辆路径构建，并结合Tent混沌初始化、局部搜索策略提升算法搜索能力。以江浙沪区域1个配送中心和40个客户点为算例进行验证，结果表明，算法最终启用5辆冷藏车完成全部配送任务，包括3辆中型冷藏车和2辆小型冷藏车。最终配送总成本为15757.04元，总配送里程为3149.39 km，时间惩罚成本为41.82元，客户满意度为0.9649。研究结果表明，所构建模型能够根据客户需求、空间分布和车型成本差异自动完成车辆选择与路径优化，在降低综合配送成本的同时保持较高客户满意度，可为生鲜农产品企业多车型配送组织与路径决策提供参考。

Abstract: Against the backdrop of continuously growing consumer demand for fresh agricultural products and increasingly stringent requirements for refined cold chain logistics operations, enterprise delivery activities face challenges such as high delivery costs, insufficient vehicle resource matching, unreasonable route arrangements, and difficulty in simultaneously maintaining customer service levels. Addressing the practical characteristics of multi-vehicle collaboration, dual constraints of weight and volume, and time window service requirements in fresh agricultural product delivery, this paper constructs a multi-vehicle route optimization model considering transportation costs, vehicle fixed costs, refrigeration costs, cargo damage costs, time penalty costs, and customer satisfaction. An improved Gray Wolf algorithm is designed to solve this model. The algorithm uses random key encoding to represent the customer access order and vehicle activation order, implements customer allocation and vehicle route construction through constraint decoding, and enhances the algorithm’s search capability by incorporating Tent chaotic initialization and local search strategies. The algorithm was validated using a distribution center and 40 customer locations in the Jiangsu, Zhejiang, and Shanghai region as a case study. Results show that the algorithm ultimately utilizes 5 refrigerated trucks to complete all delivery tasks, including 3 medium-sized and 2 small refrigerated trucks. The total delivery cost was 15757.04 yuan, the total delivery distance was 3149.39 km, the time penalty cost was 41.82 yuan, and the customer satisfaction rate was 0.9649. The results demonstrate that the constructed model can automatically select vehicles and optimize routes based on customer needs, spatial distribution, and vehicle cost differences, maintaining high customer satisfaction while reducing overall delivery costs. This model can provide a reference for multi-vehicle delivery organization and route decision-making for fresh agricultural product enterprises.