带时间窗的物流配送区域划分模型及其算法

Logistics distribution region partition model with time windows and its algorithms

针对烟草行业存在客户点大规模、客户需求量不固定、配送车辆最大行驶距离限制以及客户点送货时间不固定等特点,综合考虑配送的多产品、多客户、时间限制等影响因素,通过聚类方法划分不同的配送单元,应用整数规划选择中转站不固定配送单元,以物流配送网络构建的总成本最小化为目标函数,建立了基于配送单元的固定成本和变动成本以及带时间窗的时滞成本的数学规划模型,并提出了一种改进粒子群-遗传混合算法进行直接求解.该算法在评价函数中隐含加入了距离和时间等约束条件,并设计了算法间选择性赋予方法,具有较高的全局和局部搜索能力.实例仿真表明,该混合算法的优化性能和效率优于PSO算法、GA算法、GA-PSO算法和M PSO算法,因此能够更有效地解决大规模配送点的物流配送区域划分问题.

Tobacco industry has large-scale customers,customers＇ demand is not fixed,the maximum travel distance of distribution vehicles is limited and customers＇ delivery time is not fixed.Based on these features the logistics distribution region is divided into different distribution units by cluster methods,and integer programming is applied to choose unfixed transfer stations.Finally on the basis of fixed cost and variable cost and delay cost with time windows of distribution units,a mathematical programming model is established to minimize the cost of logistics distribution network considering multi-product,multi-client,time constraints and other factors.An EPSO-GA（extended particle swarm optimization-genetic algorithms） is also presented to solve the model.In this algorithm distance and time constraints are added into evaluation function,and selective interaction between the algorithms is designed,therefore,it provides a higher global and local search capability.The simulation results show that the hybrid algorithm can solve distribution region partition problems which include large-scale distribution points.It is more effective than MPSO（multi-phases particle swarm optimization algorithm）,GA（genetic algorithms）,PSO（particle swarm optimization） and GA-PSO（genetic algorithm-particle swarm optimization）algorithms.