应急物资配送动态调度的粒子群算法  

Emergency supplies distributing and vehicle routes programming based on particle swarm optimization

作　　者：田军[1] 马文正[1] 汪应洛[1] 王刊良[1] 

机构地区：[1]西安交通大学管理学院,西安710049

出　　处：《系统工程理论与实践》2011年第5期898-906,共9页Systems Engineering-Theory & Practice

基　　金：国家自然科学基金-(70971108)

摘　　要：应急条件下的物资配送与调度面临着需求信息不准确、需求紧急程度差异和运输路网动态变化的复杂环境,借助模糊数学中的三角模糊数描述应急物资需求量,利用连续速度时间依赖函数模拟真实的动态路网交通状况,并考虑不同需求点的需求紧急程度差异,建立了针对性的应急物资配送动态调度的多目标数学模型;通过设计粒子群优化算法,采用"离散-连续向量混合编码"方案和加权整合的适应值函数导向机制,结合连续更新的位置和速度操作策略,建立了针对这类含有离散和连续变量组合的优化模型的快速高效求解算法;最后,结合两个实际的算例进行了数值实验与分析,通过与用Matlab求得的解析解的比较,证明算法收敛速度快、鲁棒性强,从而为应急条件下的物资配送动态调度提供了有效和可靠的方法.Fuzzy demands, dynamic transportation network, and prioritized supplying and distribution are the typical characters of emergency supplies under eruption incident（s）. Making use of the triangular fuzzy numbers of fuzzy mathematics to describe the emergency demands for supplies, and the speed of time-dependent function to simulate the real dynamics states of the road network, as well as thinking of the prioritized condition, the paper set up an optimization model for scheduling emergency supplies, which had a multi-objective and could choose a road to avoid congestion in special period of time, and strictly meet the needs of time requirement of the real supply orders. Then a Particle Swarm Optimization algorithm has been designed according to the characteristics of the model. A vector encoding method with ＂discrete - continuous variables encoding＂ for every particle was put forward, a Fitness Function was proposed according to the objective functions, a strategy for continues renewing the location and velocity of each particle was proposed. It could better solve the problems of discrete and continuous combinatorial optimization variables for the mathematical model of the emergency dispatch. With the help of an empirical analysis in combination with two examples, the algorithms was proved suitable and validity when the direct comparison of exact solutions obtained by the MATLAB was done.

关 键 词：应急物资配送 动态车辆路径规划 多目标优化调度 粒子群算法 

分 类 号：F252.8[经济管理—国民经济]