基于混合算法的电力物料车辆排队服务优化

Queue-to-service Optimization of Electric Power Material Vehicles Based on Hybrid Algorithm

针对电力物料产品出入库频繁,扰动大,异质物料装卸时间差异大,车辆排队服务时间不确定等问题,建立电力物料装卸排队服务优化模型,模型中考虑了转换装卸仓库的时间对优化方案的影响。利用遗传算法对优良基因的记忆保留特性和算法收敛性以及粒子群算法迭代的方向性,提出混合遗传粒子群算法,加入了Metropolis抽样准则以提高算法跳出局部最优的能力。最后对国家电网天津市电力物资公司智慧物流园区实际数据算例进行仿真优化分析,模型优化后比实际的车辆平均空闲时间、最大空闲时间和总空闲时间都有大幅度减少,表明了模型的可行性和算法的有效性。相关研究模型已在企业智慧物流园区供应链多环节协同关键技术研究项目中进行分析与验证,并应用于企业实际仓储服务信息系统,极大地减小了企业运输车辆排队等待服务现象,减少了时间与成本浪费,具有较强的实际意义与应用价值。

In an accelerating economy,the rapidly expanding operation scale of an enterprise would put excessive stress on the various material transportation nodes in a supply chain,as evidenced by the long queue of transportation vehicles waiting service in warehousing parks.Due to the great disturbance of the inbound and outbound orders,warehouses are unable to precisely estimate the time necessary to finish loading and un⁃loading the orders and the queuing time for service is also difficult to determine,leading to long vehicle queues that cause huge waste of personnel and costs.This phenomenon is especially obvious in the transportation and storage process of electric power materials.In this paper,in view of the characteristics of electric power materials such as wide variety,huge differ⁃ence in weight,size,loading/unloading time,and requirement on operation methods and tools,as well as is⁃sues including frequent inbound and outbound operations,large disturbance,vast differences in loading and unloading times across heterogeneous materials,and uncertain vehicle queue-to-service time,etc.,we estab⁃lished the queue-to-service optimization model in inbound and outbound operations of electric power mate⁃rials,which aims to minimize both the average and maximum idle time of all vehicles.The model considers the impact of the time spent on relocating the materials between the loading and unloading warehouses on the op⁃timization plan,and groups together the inbound vehicles and outbound orders arriving at the park within a dy⁃namic period of time in determining the arrival time and order of the outbound vehicles,thereby reducing the queuing time and cost of the vehicles.Next,we combined advantages of the genetic algorithm in gene memo⁃ry/retention and convergence with the iterative directionality of the particle swarm algorithm to design a hy⁃brid genetic particle swarm algorithm,and added in the Metropolis sampling criterion to enable the algorithm to jump out of local optimality.Finally,we had a simulation analysis based on the actual data of the Smart Logis⁃tics Park of State Grid Tianjin Electric Power Materials Company.The result showed that after optimization,the average idle time,maximum idle time and total idle time of the transportation vehicles were significantly re⁃duced,demonstrating the feasibility and effectiveness of the model and algorithm.The relevant research model has been analyzed and verified in the supply chain multi-link collaboration key technology research project of the enterprise's smart logistics park,and has been applied in the enterprise's warehousing service information system,which has greatly reduced the queuing behavior of the transportation vehicles,saving both time and cost,lending proof to the strong practical significance and application value of this research.