高速铁路列车连带晚点的智能预测及特征识别

张琦; 陈峰; 张涛; 袁志明

doi:10.16383/j.aas.c190188

高速铁路列车连带晚点的智能预测及特征识别

doi: 10.16383/j.aas.c190188

张琦^1,,
陈峰^1,,
张涛^1,,
袁志明^1,

1.
中国铁道科学研究院集团有限公司通信信号研究所, 北京 100081

基金项目: 国家自然科学基金(61790575), 中国国家铁路集团有限公司科技研究开发计划课题(N2019G020), 中国铁道科学研究院集团有限公司科研课题(2018YJ070, 2018YJ076)资助

详细信息

作者简介:
张琦：中国铁道科学研究院集团有限公司首席研究员. 1998年获得中国铁道科学研究院博士学位. 主要研究方向为铁路通信信号, 列车自动驾驶, 列车运行控制, 多列车智能调度与协同控制. E-mail: zhangqi@rails.cn

陈峰：中国铁道科学研究院集团有限公司通信信号研究所副研究员. 2012年获得北京交通大学博士学位. 主要研究方向为铁路通信信号, 列车运行控制, 智能调度与协同控制. 本文通信作者. E-mail: chenfeng@bjtu.edu.cn

张涛：中国铁道科学研究院集团有限公司通信信号研究所副研究员. 2015年获得中国铁道科学研究院博士学位. 主要研究方向为列车调度指挥系统和调度集中系统. E-mail: 13701193534@139.com

袁志明：中国铁道科学研究院集团有限公司研究员. 2016年获得中国铁道科学研究院博士学位. 主要研究方向为铁路运营指挥, 铁路信号控制和铁路智能调度. E-mail: zhimingyuan@hotmail.com

计量
- 文章访问数: 2698
- HTML全文浏览量: 856
- PDF下载量: 239
- 被引次数: 0
出版历程
- 收稿日期: 2019-03-20
- 录用日期: 2019-09-24
- 刊出日期: 2019-12-01

Intelligent Prediction and Characteristic Recognition for Joint Delay of High Speed Railway Trains

1.
Signal and Communication Research Institute, China Academy of Railway Sciences Corporation Limited, Beijing 100081

Funds: Supported by National Natural Science Foundation of China (61790575), Science and Technology Project of China National Railway Group Corporation Limited (N2019G020), and Science and Technology Project of China Academy of Railway Sciences Corporation Limited (2018YJ070, 2018YJ076)

摘要

摘要: 列车晚点预测及分析直接服务于高速铁路调度指挥, 是列车晚点研究的重点内容. 通过对列车晚点传播结构及传播规律的分析, 提出了一种高速铁路列车连带晚点的智能预测及特征识别方法. 首先利用列车晚点自身传播及相邻线列车晚点传播的关系, 构建基于小波神经网络的列车连带晚点递阶预测模型; 然后利用列车晚点波动的线性组合方程及其结构向量, 进行列车连带晚点影响值的量化; 最后综合连带晚点的实际值、预测值和影响值, 将晚点分为严重晚点、消散晚点、潜在晚点、一般晚点四种类型. 成渝高铁的实例数据表明, 小波神经网络的预测结果具有较高精度, 连带晚点的分类结果也比较符合实际, 能够为高速铁路列车连带晚点的运行调整提供数据支撑.
- 铁路运输 /
- 高速铁路 /
- 列车连带晚点 /
- 晚点预测 /
- 特征识别 /
- 神经网络
Abstract: The prediction and analysis of train delays is the core content of train delay research, which directly serves the dispatching command of high-speed railway. Based on the propagation structure and law of train delay analyzing, an intelligent prediction and feature recognition method for the large area joint train delay is proposed. Firstly, considering the relationship between train delay propagation and adjacent train delays propagation, a hierarchical prediction model of train delay is constructed using wavelet neural network. Secondly, the linear combination equation of train delay fluctuation and its structural vector are used to quantify the impact value of train delays. Finally, combined with the prediction value and impact value of continuous delays, the joint delay are divided into heavy delay, dissipation delay, potential delay and general delay. The example using the data of Chengdu-Chongqing high-speed railway shows that the prediction results of wavelet neural network have high precision, and the classification results of joint delay are more realistic, which can provide data support for large-area train delayed operation adjustment.
- Railway transportation /
- high-speed railway /
- train joint delay /
- delay prediction /
- feature recognition /
- neural network

HTML全文

图 1 成渝高铁4趟晚点列车的恢复过程

Fig. 1 Restoration process of 4 delayed trains on Chengdu-Chongqing high-speed railway

下载: 全尺寸图片幻灯片

图 2 同一车站相邻车次的到达晚点时间散点图

Fig. 2 Scatter plot arrival time of adjacent delayed trains at the same station

下载: 全尺寸图片幻灯片

图 3 同一列车相邻车站的到达晚点时间散点图

Fig. 3 Scatter plot arrival delayed time of adjacent stations of the same train

下载: 全尺寸图片幻灯片

图 4 小波神经网络的拓扑结构图

Fig. 4 Topology of wavelet neural networks

下载: 全尺寸图片幻灯片

图 5 晚点分类的Moran散点示意图

Fig. 5 Moran scatter plot for train delayed classification

下载: 全尺寸图片幻灯片

图 6 晚点时间实际值与预测值的对比

Fig. 6 Comparison of actual and predicted train delay time

下载: 全尺寸图片幻灯片

图 7 3种预测方法的训练曲线

Fig. 7 Training curves of three forecasting methods

下载: 全尺寸图片幻灯片

图 8 连带晚点的严重程度分级

Fig. 8 Severity grading of joint delayed train

下载: 全尺寸图片幻灯片

表 1 列车连带晚点的原始数据

Table 1 Raw data on joint delayed train

序号	X₁ (min)	X₂ (min)	X₃ (min)	X₄ (min)	X₅ (min)	X₆	X₇	Y (min)
1	25	1.23	0.56	22	4.3	1	1	22
2	7	2.37	0.09	7	5.7	1	1	6
3	8	1.66	0.79	9	4.8	1	2	5
4	8	1.85	0.49	7	5.5	1	1	7
5	8	2.16	0.68	10	3.8	1	1	8
6	11	2.24	0.48	11	6.2	2	1	9
$\vdots $	$\vdots $	$\vdots $	$\vdots $	$\vdots $	$\vdots $	$\vdots $	$\vdots $	$\vdots $

下载: 导出CSV

表 2 4种预测方法的误差比较

Table 2 Error comparison of four prediction methods

预测模型	MAE (min)	MRE (%)	AAE (min)	ARE (%)
多元线性回归	11.38	51.71	4.29	25.89
随机森林	5.92	20.60	2.28	11.82
BP神经网络	6.11	29.54	2.42	14.50
小波神经网络	3.67	14.16	1.46	8.01

下载: 导出CSV

参考文献(27)

[1]	孟令云, Goverde R M P. 基于实际数据分析的列车晚点传播过程构建方法与实例. 北京交通大学学报, 2012, 36(6): 15−20 doi: 10.3969/j.issn.1673-0291.2012.06.003 1 Meng Ling-Yun, Goverde R M P. A method for constructing train delay propagation process by mining train record data. Journal of Beijing Jiaotong University, 2012, 36(6): 15−20 doi: 10.3969/j.issn.1673-0291.2012.06.003
[2]	周方明, 程先东, 谢美全, 张增勇, 毛保华. 地铁列车初始布点的鲁棒性研究. 物流技术, 2010, 29(8): 47−51 2 Zhou Fang-Ming, Cheng Xian-Dong, Xie Mei-Quan, Zhang Zeng-Yong, Mao Bao-Hua. Research on the robustness of the initial distribution of metro trains. Logistics Technology, 2010, 29(8): 47−51
[3]	3 Ho T K, Norton J P, Goodman C J. Optimal traffic control at railway junctions. IEE Proceedings-Electric Power Applications, 1997, 144(2): 140−148 doi: 10.1049/ip-epa:19970941
[4]	4 Fay A. A fuzzy knowledge-based system for railway traffic control. Engineering Applications of Artificial Intelligence, 2000, 13(6): 719−729 doi: 10.1016/S0952-1976(00)00027-0
[5]	5 Mazzarello M, Ottaviani E. A traffic management system for real-time traffic optimisation in railways. Transportation Research. Part B, 2007, 41(2): 246−274 doi: 10.1016/j.trb.2006.02.005
[6]	胡思继, 孙全欣, 胡锦云, 杨肇夏. 区段内列车晚点传播理论的研究. 中国铁道科学, 1994, 15(2): 41−54 6 Hu Si-Ji, Sun Quan-Xin, Hu Jin-Yun, Yang Zhao-Xia. Research on theories of train delay propagation in a railway district. China Railway Science, 1994, 15(2): 41−54
[7]	7 Meester L E, Muns S. Stochastic delay propagation in railway networks and phase-type distributions. Transportation Research, Part B (Methodological), 2007, 41(2): 218−230 doi: 10.1016/j.trb.2006.02.007
[8]	8 Goverde R M P. A delay propagation algorithm for large-scale railway traffic networks. Transportation Research Part C Emerging Technologies, 2010, 18(3): 269−287 doi: 10.1016/j.trc.2010.01.002
[9]	9 Yuan J, Hansen I A. Optimizing capacity utilization of stations by estimating knock-on train delays. Transportation Research, Part B (Methodological), 2007, 41(2): 202−217 doi: 10.1016/j.trb.2006.02.004
[10]	黄平, 彭其渊, 文超, 杨宇翔. 武广高速铁路列车晚点恢复时间预测的随机森林模型. 铁道学报, 2018, 40(7): 1−9 10 Huang Ping, Peng Qi-Yuan, Wen Chao, Yang Yu-Xiang. Random forest prediction model for Wuhan-Guangzhou HSR primary train delays recovery. Journal of the China Railway Society, 2018, 40(7): 1−9
[11]	11 Zhuang H, Feng L P, Wen C, Peng Q Y, Tang Q Z. High-speed railway train timetable conflict prediction based on fuzzy temporal knowledge reasoning. Engineering, 2016, 2(3): 366−373 doi: 10.1016/J.ENG.2016.03.019
[12]	12 Burdett R L, Kozan E. A sequencing approach for creating new train timetables. OR Spectrum, 2010, 32(1): 163−193
[13]	柏赟, 何天健, 毛保华. 一种交叉线干扰情形下列车晚点恢复运行控制方法. 交通运输系统工程与信息, 2011, 11(5): 114−122 doi: 10.3969/j.issn.1009-6744.2011.05.017 13 Bai Yun, Ho Tin-Kin, Mao Bao-Hua. Train control to reduce delays upon service disturbances at railway junctions. Journal of Transportation Systems Engineering and Information Technology, 2011, 11(5): 114−122 doi: 10.3969/j.issn.1009-6744.2011.05.017
[14]	季学胜, 孟令云. 列车到发时刻与进路同步优化的高速铁路列车运行调整模型. 中国铁道科学, 2014, 35(4): 117−123 doi: 10.3969/j.issn.1001-4632.2014.04.17 14 Ji Xue-Sheng, Meng Ling-Yun. Train operation adjustment model for synchronously optimizing train arrival/departure time and route on high speed railway network. China Railway Science, 2014, 35(4): 117−123 doi: 10.3969/j.issn.1001-4632.2014.04.17
[15]	袁志明, 张琦, 黄康, 冯姗姗. 基于随机森林的列车到站时间预测方法. 铁道运输与经济, 2016, 38(5): 60−63, 79 15 Yuan Zhi-Ming, Zhang Qi. Huang Kang. Feng Shan-Shan. Forecast method of train arrival time based on random forest algorithm. Railway Transport and Economy, 2016, 38(5): 60−63, 79
[16]	孙略添, 宋瑞, 何世伟, 殷玮川. 技术站列车晚点时间预测方法. 北京交通大学学报, 2018, 42(1): 94−98,126 doi: 10.3969/j.issn.1672-8106.2018.01.013 16 Sun Lue-Tian. Song Rui. He Shi-Wei. Yin Wei-Chuan. Prediction method of train delay time in technology service station. Journal of Beijing Jiaotong University, 2018, 42(1): 94−98,126 doi: 10.3969/j.issn.1672-8106.2018.01.013
[17]	庄河, 文超, 李忠灿, 汤轶雄, 黄平. 基于高速列车运行实绩的致因 − 初始晚点时长分布模型. 铁道学报, 2017, 39(9): 25−31 doi: 10.3969/j.issn.1001-8360.2017.09.004 17 Zhuang He, Wen Chao, Li Zhong-Can, Tang Yi-Xiong, Huang Ping. Cause based primary delay distribution models of high-speed trains on account of operation records. Journal of the China Railway Society, 2017, 39(9): 25−31 doi: 10.3969/j.issn.1001-8360.2017.09.004
[18]	刘宇, 黄凯. 基于极大代数的城际高速列车晚点传播研究. 综合运输, 2017, 39(9): 68−73, 107 18 Liu Yu, Huang Kai. Analysis of delay propagation of network using max-plus theory. China Transportation Review, 2017, 39(9): 68−73, 107
[19]	19 McCulloch W S, Pitts W. A logical calculus of the ideas immanent in nervous activity. The Bulletin of Mathematical Biophysics, 1943, 5(4): 115−133 doi: 10.1007/BF02478259
[20]	诸静. 智能预测控制及其应用. 浙江大学出版社, 2002. 130−137 Zhu Jing. Intelligent Predictive Control and Its Application. Zhejiang University Press, 2002. 130−137
[21]	朱大奇. 人工神经网络研究现状及其展望. 江南大学学报(自然科学版), 2004, 3(1): 103−110 doi: 10.3969/j.issn.1671-7147.2004.01.027 21 Zhu Da-Qi. The research progress and prospects of artificial neural networks. Journal of Southern Yangtze University (Natural Science Edition, 2004, 3(1): 103−110 doi: 10.3969/j.issn.1671-7147.2004.01.027
[22]	22 Grossmann A, Kronland-Martinet R, Morlet J. Reading and understanding continuous wavelet transforms. Wavelets, 1990, 31(9): 2−20
[23]	吕柏权, 李天铎, 吕崇德, 刘兆辉. 一种用于函数学习的小波神经网络. 自动化学报, 1998, 24(4): 548−551 23 Lv Bai-Quan, Li Tian-Duo, Lv Chong-De. Liu Zhao-Hui. Wavelet neural network for function learning. Acta Automatica Sinica, 1998, 24(4): 548−551
[24]	王群仙, 李少远, 李俊芳. 小波分析及其在控制中的应用. 控制与决策, 2000, 15(4): 385−389, 394 doi: 10.3321/j.issn:1001-0920.2000.04.001 24 Wang Qun-Xian, Li Shao-Yuan, Li Jun-Fang. Wavelet analysis and its applications in control. Control and Decision, 2000, 15(4): 385−389, 394 doi: 10.3321/j.issn:1001-0920.2000.04.001
[25]	王正武, 黄中祥. 短时交通流预测模型的分析与评价. 系统工程, 2003, 21(6): 97−100 doi: 10.3969/j.issn.1001-4098.2003.06.020 25 Wang Zheng-Wu, Huang Zhong-Xiang. An analysis and discussion on short-term traffic flow forecasting. Systems Engineering, 2003, 21(6): 97−100 doi: 10.3969/j.issn.1001-4098.2003.06.020
[26]	王嵘冰, 徐红艳, 李波, 冯勇. BP神经网络隐含层节点数确定方法研究. 计算机技术与发展, 2018, 28(4): 37−41 26 Wang Rong-Bing, Xu Hong-Yan, Li Bo, Feng Yong. Research on method of determining hidden layer nodes in BP neural network. Computer Technology and Development, 2018, 28(4): 37−41
[27]	陈绍宽, 韦伟, 毛保华, 关伟. 基于改进时空Moran's I指数的道路交通状态特征分析. 物理学报, 2013, 62(14): 1−7 27 Chen Shao-Kuan, Wei Wei, Mao Bao-Hua, Guan Wei. Analysis on urban traffic status based on improved spatio-temporal Moran's I. Acta Physica Sinica, 2013, 62(14): 1−7