高铁故障晚点时间预测的支持向量回归模型*

doi:10.16265/j.cnki.issn1003-3033.2019.S2.003

中国安全科学学报 ›› 2019, Vol. 29 ›› Issue (S2): 18-23.doi: 10.16265/j.cnki.issn1003-3033.2019.S2.003

高铁故障晚点时间预测的支持向量回归模型^*

汤轶雄¹, 徐传玲¹, 文超^*1,2,3 副教授, 李忠灿¹, 宋邵杰³

1 西南交通大学交通运输与物流学院,四川成都 610031;
2 综合交通运输智能化国家地方联合工程实验室,四川成都 610031;
3 中国铁路广州局集团有限公司调度所,广东广州 510088

收稿日期:2019-08-05 修回日期:2019-10-05 出版日期:2019-12-30 发布日期:2020-10-28
通讯作者: ** 文超(1984—),男,江西宜春人,博士,副教授,博士生导师,主要从事铁路运输组织优化、交通大数据应用等方面的研究。E-mail: wenchao@swjtu.cn。
作者简介:汤轶雄 (1994—),男,江西宜春人,博士研究生,主要研究方向为高速铁路晚点管理。E-mail: tangyixiong@my.swjtu.edu.cn。
基金资助:
国家自然科学基金资助(71871188, 61503311);四川省科技厅应用基础研究项目(2018JY0567)。

Support vector regression models for delay time predicting considering high-speed rail facility failure

TANG Yixiong¹, XU Chuanling, WEN Chao^1,2,3, LI Zhongcan¹, SONG Shaojie³

1 School of Transportation & Logistics, Southwest Jiaotong University, Chengdu Sichuan 610031, China;
2 National United Engineering Laboratory of Integrated and Intelligent Transportation, Chengdu Sichuan 610031, China;
3 Dispatching Department, China Railway Guangzhou Group Co., Ltd, Guangzhou Guangdong 510088, China

Received:2019-08-05 Revised:2019-10-05 Online:2019-12-30 Published:2020-10-28

摘要/Abstract

摘要： 为准确预测高速铁路(HSR)列车故障引起的总晚点时间,基于广铁集团高速铁路列车晚点实绩数据,分别运用软间隔支持向量机回归(ε-SVR)和容错比支持向量机回归(ν-SVR)方法以初始晚点时间、影响列车数、晚点致因为自变量,总晚点时间为因变量构建SVR模型。使用测试数据进行模型预测能力评估,结果表明:在20%相对允许误差范围内,ε-SVR和ν-SVR模型的预测精度均超过了0.8,且ν-SVR模型的预测精度要高于ε-SVR模型。

关键词: 高速铁路(HSR), 列车运行实绩, 初始晚点, 晚点时间预测, 支持向量回归(SVR)模型

Abstract: In order to accurately predict the total delay time caused by train fault, the support vector regression model was established by soft margin SVR method (ε-SVR) and fault tolerance ratio SVR (ν-SVR) method on the basis of real operation data of HSR obtained from Guangzhou Railway Bureau. The primary delay time, the number of affected trains and delay causes were set as the independent variables while the total influenced time was the dependent variable. The predictive ability assessment was carried out using the testing data, and the results show that the predicting accuracies of both ε-SVR and ν-SVR model are over 0.8 when the permitted error was within 20%, and that ν-SVR model has a higher predicting accuracy than ε-SVR model.

Key words: high-speed rail, train real operation data, primary delay, delay time prediction, support vector regression (SVR) model

中图分类号:

X951

汤轶雄, 徐传玲, 文超, 李忠灿, 宋邵杰. 高铁故障晚点时间预测的支持向量回归模型^*[J]. 中国安全科学学报, 2019, 29(S2): 18-23.

TANG Yixiong, XU Chuanling, WEN Chao, LI Zhongcan, SONG Shaojie. Support vector regression models for delay time predicting considering high-speed rail facility failure[J]. China Safety Science Journal, 2019, 29(S2): 18-23.

参考文献 13

[1]	OLARU D, SMITH B. Modelling behavioural rules for daily activity scheduling using fuzzy logic[J]. Transportation, 2005, 32 (4): 423-441.
[2]	BOUCH C J, ROBERTS C. State of the art for European track maintenance and renewal logistics[J]. Proceedings of the Institution of Mechanical Engineers, Part F: Journal of Rail and Rapid Transit, 2010, 224(4):319-326.
[3]	RADTKE A, HAUPTMANN D. Automated planning of timetables in large railway networks using a microscopic data basis and railway simulation techniques[J]. WIT Transactions on The Built Environment, 2004, 7:615-625.
[4]	CONTE C. Identifying dependencies among delays[D]. Göttingen: University of Göttingen, 2007.
[5]	HANSEN I A, GOVERDE R M, DIRK J M. Online train delay recognition and running time prediction[C]. 13th International IEEE Conference on. IEEE, 2010: 1 783-1 788.
[6]	BERGER A, GEBHARDT A, MÜLLER-HANNEMANN M, et al. Stochastic delay prediction in large train networks[C]. 11th Workshop on Algorithmic Approaches for Transportation Modelling, Optimization, and Systems, 2011:100-111.
[7]	MILINKOVIĆ S, MILAN M, SLAVKO V, et al. A fuzzy Petri net model to estimatetrain delays[J]. Simulation Modelling Practice and Theory, 2013, 33:144-157.
[8]	KECMAN P, GOVERDE R M. Online data-driven adaptive prediction of train event times[J]. IEEE Transactions on Intelligent Transportation Systems, 2015, 16(1): 465-474.
[9]	WALLANDER J, MÄKITALO M. Data mining in rail transport delay chain analysis[J]. International Journal of Shipping and Transport Logistics, 2012, 4(3): 269-285.
[10]	YAGHINI M, KHOSHRAFTAR M M, SEYEDABADI M. Railway passenger train delay prediction via neural network model[J]. Journal of Advanced Transportation, 2013,47 (3):355-368.
[11]	MARKOVIĆ N, SANJIN M, KONSTANTIN S, et al. Analyzing passenger train arrival delays with support vector regression[J]. Transportation Research Part C: Emerging Technologies, 2015, 56: 251-262.
[12]	黄平,彭其渊,文超,等.高速铁路故障分类及其影响列车数模型[J].中国安全科学学报,2018,28(增2):46-53.HUANG Ping, PENG Qiyuan, WEN Chao, et al. Study on high-speed railway disruption classification and model of its influence on train number[J]. China Safety Science Journal, 2018,28(S2):46-53.
[13]	黄平,李忠灿,文超,等.高速铁路故障时空分布及持续时长分布特征研究[J].中国安全科学学报,2018,28(增2):99-104.HUANG Ping, LI Zhongcan, WEN Chao, et al. Study on spatial-temporal and duration distribution characteristics of high-speed railway disruptions[J]. China Safety Science Journal, 2018,28(S2):99-104.

高铁故障晚点时间预测的支持向量回归模型^*

Support vector regression models for delay time predicting considering high-speed rail facility failure

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献 13

相关文章 3

编辑推荐

Metrics

本文评价

[1]	文超, 李忠灿, 黄平, 田锐, 牟玮玮, 李力. 数据驱动的列车晚点传播研究^*[J]. 中国安全科学学报, 2019, 29(S2): 1-9.
[2]	黄平, 文超, 李忠灿, 杨宇翔, 彭其渊. 高速铁路列车晚点时间实时预测的神经网络模型^*[J]. 中国安全科学学报, 2019, 29(S1): 20-26.
[3]	文超, 杨雄, 黄平, 李忠灿, 唐优华. 铁路列车运行冲突检测与消解理论研究综述^*[J]. 中国安全科学学报, 2018, 28(S2): 66-73.