基于SOM的物流配送路径优化方法研究  

Study on routing optimization method of logistics distribution based on SOM

作　　者：李尧[1,2,3] 芮小平[3] 

机构地区：[1]内江师范学院计算机与信息科学系,四川内江641112 [2]内江师范学院网络应用项目开发重点实验室,四川内江641112 [3]中国科学院研究生院资环学院,北京100049

出　　处：《计算机工程与设计》2009年第7期1671-1673,共3页Computer Engineering and Design

基　　金：四川省应用基础重点项目(07JY29-124);四川省教育厅重点项目(2006A145)

摘　　要：合理的配送路线可以提高物流配送的效率,利用SOM网络训练过程中能够保持神经元空间拓扑特性的特点求解物流配送的最优路径。在SOM网络中以任意配送节点的坐标作为输入层,以2倍于配送节点的序列为输出层。训练时,每次从路径节点中随机选取一个节点作为输入层,通过训练保证输出层的某个点与该输入点的距离最短,最终使得每个配送节点在输出层上都有与之最近的节点相对应。给出了求解最优路径的步骤,并分别针对不同的配送节点进行实验。结果表明,基于SOM最优路径算法能够以较快的收敛速度得到最优解。Rational delivery route can improve the efficiency of logistics distribution, SOM neural network can keep the topological characteristics between the neurons, so SOM is used to solve the shortest delivery route of logistics. One delivery node is selected as input vector of the network randomly and the out put nodes of the network is 2 times than the number of delivery nodes. When training, the network will identify the weight vector closest to the delivery node and move it and its neighbors closer to the delivery node. After training, every delivery node has the closest node respectively. The steps using SOM to find the shortest route is brought out. Test our method with different delivery nodes, results show that SOM algorithm can obtain faster convergence rate and get the better optimal solution.

关 键 词：配送路线 SOM 物流 最短路径 神经网络 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论] F713.36[自动化与计算机技术—计算机科学与技术]