[1]纪宇宣,蒋秋华,朱颖婷.基于随机距离预测的高铁客流需求研究[J].计算机技术与发展,2022,32(05):189-194.[doi:10. 3969 / j. issn. 1673-629X. 2022. 05. 032]
　JI Yu-xuan,JIANG Qiu-hua,ZHU Ying-ting.Research on Passenger Flow Demand of High-speed Rail Based on Random Distance Prediction[J].,2022,32(05):189-194.[doi:10. 3969 / j. issn. 1673-629X. 2022. 05. 032]
点击复制

基于随机距离预测的高铁客流需求研究()
分享到：

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:
32
期数:
2022年05期
页码:
189-194
栏目:
应用前沿与综合
出版日期:
2022-05-10

文章信息/Info

Title:
Research on Passenger Flow Demand of High-speed Rail Based on Random Distance Prediction
文章编号:
1673-629X(2022)05-0189-06
作者:
纪宇宣1 ; 蒋秋华2; 3 ; 朱颖婷2; 3
1. 中国铁道科学研究院研究生部,北京 100081;
2. 中国铁道科学研究院集团有限公司 电子计算技术研究所,北京 100081;
3. 北京经纬信息技术有限公司,北京 100081
Author(s):
JI Yu-xuan1 ; JIANG Qiu-hua2; 3 ; ZHU Ying-ting2; 3
1. Postgraduate Department,China Academy of Railway Sciences,Beijing 100081,China;
2. Institute of Computing Technologies,China Academy of Railway Science Corporation Limited,Beijing 100081,China;
3. Beijing Jingwei Information Technology Co. ,Ltd. ,Beijing 100081,China
关键词:
客运; 余票查询; 起讫点; 随机距离预测; 神经网络; 聚类
Keywords:
passenger transport; remaining ticket inquiry; origin-destination; random distance prediction; neural network; clustering
分类号:
TP39
DOI:
10. 3969 / j. issn. 1673-629X. 2022. 05. 032
摘要:
历史客运量与客运需求存在差距,基于余票查询数据的起讫点( OD) 客流特征分析可以较为实时地反映客运需求。 对于一些客流特征的挖掘目前主要的方法是利用聚类算法进行群体划分,进而发现每个类别的特征。 针对余票查询数据维度高,直接使用聚类算法鲁棒性较差的问题,提出了一种基于随机距离预测的高层特征抽取模型 RDP 与 K-means结合的 OD 客流聚类分析方法。 以京沪高速铁路预售期内余票查询量数据为原始数据,将乘车日期作为预分类条件,运用RDP 算法提取预分类后数据的重构特征,然后通过 K-means 算法对重构特征进行聚类。 实验结果表明,RDP K-means 算法在 Calinski-Harabaz 指数、轮廓系数、戴维森堡丁指数三种内部聚类评价指标下效果均优于传统的 K-means、PCA K-means、层次聚类、DBSCAN 等算法,证明了 RDP K-means 算法在基于余票查询数据的 OD 客流特征分析研究中的有效性,能够更好地进行 OD 类别划分、客流出行特征分析、热门 OD 挖掘,为改善运力调整等相关业务提供一定的参考依据。
Abstract:
There is a gap between the historical passenger volume and the passenger demand. The origin -destination ( OD) passenger flow characteristic analysis based on the remaining ticket query data can reflect the passenger demand in real time. For the mining of some characteristics of passenger flow,the main method is to use clustering to divide groups,and then find the characteristics of each category. In order to solve the problem that the dimension of the remaining ticket query data is high and the robustness of clustering algorithm is poor,we propose a clustering analysis method of OD passenger flow based on RDP,a high -level feature extraction model based on random distance prediction,and K-means. Taking the remaining ticket query data of Beijing Shanghai high-speed railway in the pre-sale period as the original data,the original data is pre classified according to the travel date,and the reconstructed features of the preclassified data are extracted by RDP algorithm,and then the reconstructed features are clustered by K-means algorithm. The experimental results show that the RDP K - means algorithm is better than the traditional K - means, PCA K - means, hierarchical clustering and DBSCAN algorithm in Calinski Harabaz index, contour coefficient and davisenberg index, which is proved that the RDP K - means algorithm is effective in the analysis of OD passenger flow characteristics based on the remaining ticket query data. It can better classify OD categories,analyze passenger flow characteristics and mine popular OD,which provides a certain reference for improving transport capacity adjustment and other related businesses.
更新日期/Last Update: 2022-05-10