A train running time prediction model based on domestic and foreign railway operation data

doi:10.16265/j.cnki.issn1003-3033.2022.06.2634

Abstract

Abstract:

In order to accurately predict train running time while considering requirement on accuracy and timeliness of its operation, a time prediction model was established. Firstly, data of railway timetable were extracted and analyzed to determine possible influencing factors of its running time, and Box-Cox transformation was used to normalize operation data considering their non-normality. Then, the model's input features and hyper-parameters were optimized based on decision tree and grid-search algorithm, respectively, and its performance was improved. Finally, a train running time prediction model was established by adopting HGBT based on optimized input features and hyper-parameters, and operation data of a Chinese and European railway were utilized to evaluate works at each stage. The results show that Box-Cox transformation can significantly improve data normality and goodness-of-fit of the prediction model, while grid-search algorithm can simultaneously improve the model's efficiency and accuracy. Compared with other commonly used machine learning algorithms for train running time prediction, the proposed HGBT model features higher accuracy and efficiency.

Key words: railway operation data, train running time, prediction model, Box-Cox transformation, decision tree, histogram-based gradient boosting tree (HGBT)

TANG Tao, GAN Jing. A train running time prediction model based on domestic and foreign railway operation data[J]. China Safety Science Journal, 2022, 32(6): 123-130.

Figures/Tables 10

Fig.1

Fig.2

Tab.1

Fig.3

Fig.4

Tab.2

Fig.5

Tab.3

Fig.6

Fig.7

References 16

[1]	FISCHETTI M, MONACI M. Using a general-purpose mixed-integer linear programming solver for the practical solution of real-time train rescheduling[J]. European Journal of Operational Research, 2017, 263(1): 258-264. doi: 10.1016/j.ejor.2017.04.057
[2]	何华武, 朱亮, 李平, 等. 智能高铁体系框架研究[J]. 中国铁路, 2019 (3): 1-8.
	HE Huawu, ZHU Liang, LI Ping, et al. Research on intelligent high speed rail system framework[J]. China Railway, 2019(3): 1-8.
[3]	胡瑞, 徐传玲, 冯永泰, 等. 广深高速铁路列车分类型晚点预测[J]. 中国安全科学学报, 2019, 29(增2): 181-186.
	HU Rui, XU Chuanling, FENG Yongtai, et al. Prediction of train delay of Guangzhou Shenzhen high speed railway by type[J]. China Safety Science Journal, 2019, 29(S2): 181-186. doi: 10.16265/j.cnki.issn1003-3033.2019.S2.030
[4]	BKER T, SEYBOLD B. Stochastic modelling of delay propagation in large networks[J]. Journal of Rail Transport Planning & Management, 2012, 2(1/2): 34-50.
[5]	CORMAN F, KECMAN P. Stochastic prediction of train delays in real-time using Bayesian networks[J]. Transportation Research Part C: Emerging Technologies, 2018, 95: 599-615. doi: 10.1016/j.trc.2018.08.003
[6]	胡雨欣, 彭其渊, 鲁工圆, 等. 基于初始晚点和冗余时间的列车晚点恢复时间预测模型[J]. 交通运输工程与信息学报, 2020, 18(2):93-102.
	HU Yuxin, PENG Qiyuan, LU Gongyuan, et al. Prediction model of train delay recovery time based on initial delay and redundant time[J]. Journal of Transportation Engineering and Information, 2020, 18(2):93-102.
[7]	龙时丹. 面向高速铁路的全自动运行系统应用[J]. 中国安全科学学报, 2020, 30(增1): 92-97.
	LONG Shidan. Application of automatic operation system for high-speed railway[J]. China Safety Science Journal, 2020, 30(S1): 92-97. doi: 10.16265/j.cnki.issn1003-3033.2020.S1.017
[8]	NAIR R, HOANG T L, LAUMANNS M, et al. An ensemble prediction model for train delays[J]. Transportation Research Part C: Emerging Technologies, 2019, 104: 196-209. doi: 10.1016/j.trc.2019.04.026
[9]	孙延浩, 张琦, 任禹谋, 等. 基于改进灰色马尔科夫模型的列车晚点预测[J]. 计算机仿真, 2020, 37(6):117-121,135.
	SUN Yanhao, ZHANG Qi, REN Yumou, et al. Prediction of train delay based on improved grey Markov model[J]. Computer Simulations, 2020, 37(6):117-121,135.
[10]	胡瑞, 文超, 张梦颖, 等. 高速列车晚点预测的机器学习模型[J]. 中国铁路, 2020(11):72-77.
	HU Rui, WEN Chao, ZHANG Mengying, et al. Machine learning model for high speed train delay prediction[J]. China Railway, 2020(11):72-77.
[11]	张琦, 陈峰, 张涛, 等. 高速铁路列车连带晚点的智能预测及特征识别[J]. 自动化学报, 2019, 45(12): 2251-2259.
	ZHANG Qi, CHEN Feng, ZHANG Tao, et al. Intelligent prediction and feature recognition of train associated delay in high speed railway[J]. Acta Automatica Sinica, 2019, 45(12): 2251-2259.
[12]	廖璐, 陈峰, 葛晓程, 等. 基于GBDT的列车晚点时长预测模型研究[J]. 铁道标准设计, 2022, 65(8): 149-154,176.
	LIAO Lu, CHEN Feng, GE Xiaocheng, et al. Research on prediction model of train delay based on GBDT[J]. Railway Standard Design, 2022, 65(8): 149-154,176.
[13]	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. doi: 10.1002/atr.193
[14]	MARKOVIĆ N, MILINKOVIĆ S, TIKHONOV K S, et al. Analyzing passenger train arrival delays with support vector regression[J]. Transportation Research Part C: Emerging Technologies, 2015, 56: 251-262. doi: 10.1016/j.trc.2015.04.004
[15]	HUANG Ping, WEN Chao, FU Liping, et al. A hybrid model to improve the train running time prediction ability during high-speed railway disruptions[J]. Safety Science, 2020:Doi: 10.1016/j.ssci.2019.104510. doi: 10.1016/j.ssci.2019.104510
[16]	KE Guolin, MENG Qi, FINLEY T, et al. Lightgbm: a highly efficient gradient boosting decision tree[J]. Advances in Neural Information Processing Systems, 2017, 30:3146-3154.

指标	转换前	转换后
S	1.293	0.001
V	3.139	0.062

评估指标	Box-Cox 转换数据	未经Box-Cox 转换数据
R²	0.912	0.908
MAE/min	0.764	0.779
MAPE/%	5.59	6.17

线路	指标	HGBT	RF	AB	GB	ANN	SVM
C	MAE	0.764	0.771	1.145	0.793	0.907	1.238
	MAPE/%	5.59	5.62	8.69	5.81	6.72	8.77
	R²	0.912	0.902	0.864	0.908	0.893	0.812
E	MAE	0.356	0.371	0.392	0.376	0.402	0.756
	MAPE/%	7.54	7.78	7.90	8.18	8.94	19.42
	R²	0.876	0.869	0.862	0.872	0.868	0.766