基于多目标优化的糖厂运输车调度

Multi-objective Optimization Approach to Transportation Vehicle Scheduling in Sugar Mill

摘要:

糖厂是甘蔗产业链的核心枢纽,糖厂的高效运转对于保障甘蔗从田间到厂区的顺畅运输与稳定供应、实现原料高效加工具有重要意义。针对榨季常见的车辆排队、司机收入不均等问题,本研究将实际调度过程抽象为排队卸蔗的多行程车辆路径问题(Multi-trip vehicle routing problem with unloading queue,MTVRP-UQ)。在此基础上,通过优化运输订单分配方式,构建了旨在最小化运输车排队等待时间和最小化运输车司机收入差异的多目标优化模型。为有效求解该模型,提出一种多策略改进的多目标遗传算法(Multi-strategy enhanced multi-objective genetic algorithm,MS-MOGA)。通过Sigmoid函数的动态交叉、变异率,提高搜索稳定性;集成大邻域搜索作为局部强化算子,有效提升模型解的质量;采用基于世代距离的自适应变异策略,避免算法陷入早熟收敛。以广西崇左南华糖厂及其辐射范围内的原料供应点为例,利用高德地图提取实际路网信息。基于真实地理数据集进行仿真试验,试验结果表明,与标准多目标遗传算法相比,MS-MOGA在等待时间和收入差异2个核心指标上分别降低5.29%与35.09%,与其他多目标优化算法相比也表现出更好的综合性能,可为糖厂榨季运输调度提供更加高效的公平性选择。

Abstract:

Sugar mills serve as the core hub of the sugarcane industry chain. Their efficient operation is crucial for ensuring smooth transportation and stable supply of sugarcane from fields to mills,as well as achieving efficient raw material processing. Addressing common issues during the crushing season such as vehicle queuing and uneven driver income,the actual scheduling process was abstracted into a multi-trip vehicle routing problem with unloading queue (MTVRP-UQ). Building upon this,a multi-objective optimization model was constructed to minimize both vehicle queueing wait times and income disparities among drivers by optimizing transport order allocation. To effectively solve this model,a multi-strategy enhanced multi-objective genetic algorithm (MS-MOGA) was proposed. Dynamic crossover and mutation rates based on the Sigmoid function enhanced search stability;integrated large neighborhood search as a local reinforcement operator to enhance solution quality;and adopted an adaptive mutation strategy based on generation distance to prevent premature convergence. Using the Nanhua Sugar Factory in Chongzuo,Guangxi,and its surrounding raw material supply points as a case study,actual road network information was extracted from AutoNavi Maps. Simulation experiments conducted on the real geographic dataset demonstrated that compared with the standard multi-objective genetic algorithm,MS-MOGA reduced waiting time and income disparity by 5.29% and 35.09%,respectively,across two core metrics. It also exhibited superior overall performance relative to other multi-objective optimization algorithms,offering a more efficient and equitable choice for sugar mill transportation scheduling during the crushing season.