交通信息与安全Issue(1)：45-50,6.DOI:10.3963/j.issn 1674-4861.2016.01.001
基于 SOM 神经网络的综合客运枢纽分级方法研究∗
A Classification Method for Multimodal Passenger Hubs Based on SOM Neural Network
朱苍晖 1李鹏林 2王压帝1
作者信息
- 1. 交通运输部规划研究院 北京 100028
- 2. 北京交通大学交通运输学院 北京 100044
- 折叠
摘要
Abstract
In order to study the criteria for classifying multi-modal passenger hubs and to guide the construction of hubs of different importance,based on the design manual of terminal and yard of different transport modes and the trans-fer passenger volumes among the modes,total number of passengers traveling outside the region and total number of pas-sengers transferring from one mode to another are proposed as two classification criteria.The two reflect the factors large-ly determine the size of passenger hubs,which is required for supporting external travel and transfer.90 multimodal pas-senger hubs are selected as the case study.Self-organizing feature map (SOM)neural network,which is featured with un-supervised self-organization,self-learning and automatic classification and no need for testing data,is applied.A cumula-tive frequency method is used to determine the thresholds for classifying hubs and improve the Euclidean distance func-tion,which solves the problem related to the classification criteria with very strong correlation.Study results show that two selected,common classification criteria can reflect diversity of the construction size of the passenger hubs of different classes.The improved method also enhances the convergence speed and clustering accuracy of the neural network.关键词
交通规划/综合客运枢纽/级别划分/SOM 神经网络/频率累积法Key words
transportation planning/multimodal passenger hub/classification/SOM neural network/cumulative frequency method分类
交通工程引用本文复制引用
朱苍晖,李鹏林,王压帝..基于 SOM 神经网络的综合客运枢纽分级方法研究∗[J].交通信息与安全,2016,(1):45-50,6.
