摘要
寻找复杂配送网络中带有容量约束的多厢车辆优化路径(MCVRP)具有很强的现实意义.将局部搜索方法与果蝇优化算法相结合,提出混合果蝇优化算法(HFOA)来解决这一问题.在该算法中,采用随机方法构造初始可行解,利用路径吸引力概率函数创建果蝇飞行路径方案,选用最优方案更新配送网络的轨迹强度.为了扩大搜索范围、提高算法质量,使用2-OPT、交换和插入3个局部搜索方法优化果蝇群的飞行路径方案.研究发现,HFOA可以有效缩短多厢车辆的最优路径长度,从而使得混合果蝇算法能够产生较好的路径规划方案.并且,在大规模复杂网络上效果更好.基于随机网络、小世界网络或无标度网络的仿真实验发现,网络的平均密度、关键“长程链接”和网络规模对配送路径长度都会产生显著影响.
It is of great practical significance to search capacity constrained optimization multi-compartment vehicle routing problem(MCVRP) in complex distribution networks. In this paper, a hybrid fly optimization algorithm(HFOA) was proposed to solve this problem by combining the local search method with the fly optimization algorithm. In the HFOA proposed, first, the initial feasible solution of MCVRP was constructed by using the stochastic method. Next, from the initial solution, the path probability function was used to create the fruit fly path. Then, three local search methods were used to optimize the path, and the optimal solution was used to update the trajectory of the distribution network. In order to expand the search scope and improve the quality of the algorithm, three local search methods, 2-OPT, switching, and insertion, were used to optimize the path of fly swarm. It is found that the HFOA proposed can effectively shorten the optimal path of multi-compartment vehicles, producing better path planning schemes and achieving better results on large-scale complex networks. The simulation experiments based on stochastic networks, small world networks, or scale-free networks show that the average density of networks, the key "long-distance links" and the size of networks have significant effects on the length of distribution routes.
作者
王成亮
李守伟
WANG Chengliang;LI Shouwei(School of Business,Shandong Normal University,Jinan 250014 ,China)
出处
《系统管理学报》
CSSCI
CSCD
北大核心
2019年第4期708-716,共9页
Journal of Systems & Management
基金
国家社会科学基金资助项目(17BGL001)
山东省社会科学基金资助项目(16CGLJ28)
关键词
车辆路径问题
多厢车辆
果蝇优化算法
局部搜索
vehicle routing problem (VRP)
multi compartment vehicle
fruit fly optimization algorithm
local search
