摘要
对于大规模客户群体,如何高效合理地规划出网点位置,在节省物流企业配送成本的前提下提高货物的周转率和及时送达率,目前已成为快递物流系统网络优化的难点。为解决此类问题,针对某地区物流公司的客户信息,采用粒子群优化的K-means聚类算法进行快递网点选址。具体过程:首先采用手肘法评估研究区域需设立的最佳快递网点数;为改善K-means初始簇中心带来的易陷入局部最优解问题,利用粒子群优化算法对数据集进行迭代寻优,重新确定初始簇中心;最后通过K-means聚类算法在全局最优解附近空间完成聚类任务,最终得到的聚类结果代表配送区域的划分方案,聚类的簇中心即为快递网点位置。此外,利用3个评价指标对粒子群优化Kmeans聚类算法和传统K-means聚类算法进行对比分析。结果表明,结合粒子群优化算法后的聚类结果其类内数据相似度更高,类间数据的差异与距离更大,取得的聚类效果更合理。
For large-scale customer groups,how to efficiently and reasonably plan a network location and improve the turnover rate and timely delivery rate of goods on the premise of saving the distribution cost of logistics enterprises is the difficulty in the network optimization of express logistics system.The K-means clustering algorithm improved by particle swarm optimization was used to locate the express outlets based on customers’information of a regional logistics company.Firstly,the Elbow method was adopted to evaluate the optimal number of express outlets to be set up in the study area.Secondly,the particle swarm optimization algorithm was used to iteratively optimize to redetermine the initial cluster center,in order to avoid the problem of falling into the local optimal solution caused by the initial cluster center of K-means.Finally,the K-means clustering algorithm was used to complete the clustering task in the space near the global optimal solution.The final clustering result represented the division scheme of the distribution area,and the cluster center was the location of the express outlets.In addition,three evaluation indexes were used to make a comparative analysis of the particle swarm optimized K-means clustering algorithm and the traditional one.The results show that the clustering results combined with the particle swarm optimization algorithm have higher similarity of intra-class data,greater difference and distance of inter-class data,and more reasonable clustering effect.
作者
倪萌萌
李春树
刘银
NI Mengmeng;LI Chunshu;LIU Yin(School of Electronic and Electrial Engineering,Ningxia University,Yinchuan 750021,China)
出处
《宁夏工程技术》
CAS
2023年第2期181-186,192,共7页
Ningxia Engineering Technology
基金
宁夏自然科学基金项目(2020AAC3033)
宁夏大学研究生创新项目(GIP2020074)。
关键词
粒子群优化
K-MEANS聚类
快递网点
簇中心
particle swarm optimization
K-means clustering
express service outlets
cluster center