配送中心选址-库存问题的粒子群算法应用  

Application of Particle Swarm Optimization for Solving Location-inventory Problem of Distribution Centers

作　　者：王非[1,2] 张佳[1,3] 孙浩杰[1] 樊根耀[1] 

机构地区：[1]长安大学经济管理学院,陕西西安710064 [2]西安外国语大学人文地理研究所,陕西西安710061 [3]西安市地下铁道有限责任公司,陕西西安710018

出　　处：《公路交通科技》2011年第12期152-158,共7页Journal of Highway and Transportation Research and Development

基　　金：陕西省科技厅自然基金(2011JM9005);中央高校专项基金重点项目(Z1102);中央高校基金(社科)(CHDW2011JC060)

摘　　要：在LMRP模型基础上,从优化角度将配送中心建设成本设为配送中心规模的线性函数,构建基于可变建设成本的LMRP模型。依据选址模型与粒子群算法特性,设计了矩阵粒子作为粒子群启发式算法初始可行解。对已有的10节点、49节点、88节点算例进行近百次测试,确定针对LMRPVCC问题的粒子群算法参数。进而利用平均计算时间与平均质量2种指标对49节点算例进行测试,得到平均计算时间为23 s,满意解比其下界平均高出12.7%的测试结果。On the basis of location model with risk pooling （ LMRP）, considering construction cost as a linear function of the size of distribution centers, we built the location-inventory model of risk pooling based on variable cost of construction （LMRPVCC） for optimization. According to the features of particle swarm optimization and the location model, we decided to apply the particle heuristic swarm optimization to solute the problem. Using J x J matrix particles as an initial feasible solution, and using the existing 10-node, 49- node and 88-node computational instances in former papers for almost a hundred times of testing, we determined the PSO algorithm parameters for LMRPVCC. On the basis of lots of testing to 49-node instance by applying the average computing time and average quality as indicators, it is concluded that average computing time was 23 s and the satisfactory solution was 12.7% higher than the lower bound.

关 键 词：运输经济 配送中心 粒子群 选址 库存 

分 类 号：U492.1[交通运输工程—交通运输规划与管理]