The absolute mileage deviation calibration of high-speed railway track geometry dynamic inspection data serves as the basis for accurately evaluating track geometry and deeply studying the evolution laws of track geometry. Addressing the issue of insufficient calibration accuracy in current absolute mileage deviation of inspection data, this paper proposes a refined calibration method for absolute mileage of inspection data assisted by railway line curve ledger. Two Dynamic Time Warping (DTW) variant algorithms are used to achieve accurate calibration of inspection data mileage. Firstly, the Derivative Dynamic Time Warping (D-DTW) algorithm is used to coarsely match the inspection data with the railway line curve ledger data, facilitating the automatic segmentation of the inspection data. Then, the Shape Dynamic Time Warping (Shape-DTW) algorithm is used to realize the accurate alignment of the curve feature points of the superelevation channel data and the railway line curve ledger data in the segmented inspection data. Finally, the exact mileage of the curve feature points in the railway line curve ledger is mapped one-to-one to those in the superelevation channel data in the inspection data, so as to realize the fine calibration of the absolute mileage deviation in track geometry dynamic inspection data. The experimental results show that the maximum error of curve length extracted by this method is less than 0.8% on a high-speed railway track dynamic inspection data set. After calibration, the correlation index of multiple inspection data in the same railway line section outperforms that before calibration, which proves the efficacy of this method.
在相对里程偏差校准方法方面,汪鑫等[8]提出了基于多次检测数据波形匹配的里程偏差修正方法,该方法通过计算相关性系数评判里程偏差,但会产生无效匹配,只能依赖相关系数约束、误差限约束、误差变化率约束识别无效匹配。许贵阳等[9]提出了高速铁路轨道几何动态检测数据自动预处理方法,该方法采用相关性系数最大化原理对一定范围内多次检测数据进行里程修正。魏晖等[10]提出基于动态时间规整(Dynamic Time Warping,DTW)算法的轨道动静态检查数据匹配方法校正里程偏差,但该方法仅考虑静态检测数据与动态检测数据的里程偏差校准,无法构建多次动态检测数据在时间维度上的关联。余宁等[11]提出一种基于卷积神经网络的铁路曲线特征点检测算法,该方法可以同时对曲线特征点进行分类和定位,但需要大量的数据标记,实际应用较困难。
本文提出了1种基于改进DTW算法的轨道几何动态检测数据里程偏差校准方法。首先,利用线路曲线台账数据构建与实际线路里程准确对应的曲线台账里程-超高基准数据序列;然后,建立曲线台账里程-超高基准数据序列与检测数据里程-超高通道数据序列的关系,使用导数动态时间规整(Derivative Dynamic Time Warping,D-DTW)算法对检测数据里程-超高通道数据序列进行曲线特征点定位和曲线自动分段;最后,利用基于图形动态时间规整(Shape Dynamic Time Warping,Shape-DTW)算法对分段后的曲线台账里程-超高基准数据序列和检测数据里程-超高通道数据序列进行精细匹配,通过将检测数据里程-超高通道数据序列中曲线特征点的原始里程校准为曲线台账里程-超高基准数据序列中曲线特征点的准确里程,实现轨道几何动态检测数据的绝对里程偏差校准。
XUWenlong. Research on Registration and Anomaly Detection Algorithm of Track Dynamic Inspection Data of High-Speed Railway [D]. Beijing: Beijing Jiaotong University, 2021. in Chinese
QUJianjun. Study on the Track Irregularity Prediction and Decision-Aided Technology Based on TQI of Raising Speed Lines [D]. Beijing: Beijing Jiaotong University, 2011. in Chinese
[5]
ALLOTTAB, COLLAV, MALVEZZIM. Train Position and Speed Estimation Using Wheel Velocity Measurements [J]. Proceedings of the Institution of Mechanical Engineers, Part F: Journal of Rail and Rapid Transit, 2002, 216 (3): 207-225.
WANGXin, WANGPing, CHENRong, et al. Mileage Error Estimation and Correction for High-Speed Railway Track Inspection Data Based on Multiple Data Waveform [J]. Journal of the China Railway Society, 2020,42 (2): 110-116. in Chinese
QINHangyuan, LIUJinzhao, WANGWeidong, et al. An Expeditious Computational Method for Twice Correction of Mileage Deviation from Vehicle Dynamic Response Inspection Data [J]. China Railway Science, 2019, 40 (4): 112-119. in Chinese
[10]
张晨.高速铁路轨道检测数据里程偏差修正算法研究[D].成都:西南交通大学,2019.
[11]
ZHANGChen. Research on Mileage Deviation Correction Algorithm of High-Speed Railway Track Detection Data [D]. Chengdu: Southwest Jiaotong University, 2019. in Chinese
[12]
汪鑫.轨道动态不平顺检测数据的里程误差评估与修正[D].成都:西南交通大学,2019.
[13]
WANGXin. Mileage Errors Estimation and Correction Model for Track Geometry Data from Track Inspection Cars [D]. Chengdu: Southwest Jiaotong University, 2019. in Chinese
WANGXin, WANGYuan, WANGPing, et al. Mileage Error Estimation and Correction of Track Geometry Data from Track Inspection Car on High Speed Railway [J]. Railway Standard Design, 2018, 62 (7): 46-51. in Chinese
XUGuiyang, SHITianyun, LIUJinzhao, et al. Research on Automatic Preprocessing Methods for Track Geometry Inspection Data of High Speed Railway [J]. China Railway Science, 2013, 34 (6): 8-12. in Chinese
WEIHui, YANGFei, ZHUHongtao, et al. A Matching Method for Dynamic and Static Inspection Data of Track Based on Dynamic Time Warping [J]. Journal of Railway Science and Engineering, 2022, 19 (1): 78-86. in Chinese
YUNing, WEIShibin, LIYing, et al. Detection Algorithm of Railway Curve Feature Point Based on Convolutional Neural Network [J]. Railway Engineering, 2021, 61 (3): 133-137. in Chinese
[22]
KEOGHE J, PAZZANIM J. Derivative Dynamic Time Warping [C]// Proceedings of the 2001 SIAM International Conference on Data Mining. Philadelphia, PA: Society for Industrial and Applied Mathematics, 2001: 1-11.
LUOLin, ZHANGGeming, WUWangqing, et al. Control of Track Irregularity in Wheel Rail System [M]. Beijing: China Railway Publishing House, 2006. in Chinese
CHENRong, LIShuai, WANGYuan, et al. Evaluation Method of Rail Regularity State Based on Track Local Fluctuation in High Speed Railway [J]. Journal of the China Railway Society, 2017, 39 (2): 105-111. in Chinese
[28]
NDOYE, M, BARKERA M, KROGMEIERJ V, et al. A Recursive Multiscale Correlation-Averaging Algorithm for an Automated Distributed Road-Condition-Monitoring System [J]. IEEE Transactions on Intelligent Transportation Systems, 2011, 12 (3): 795-808.
[29]
XUP, LIUR K, SUNQ X, et al. A Novel Short-Range Prediction Model for Railway Track Irregularity [J]. Discrete Dynamics in Nature and Society, 2012, 2012 (1): 1951-1965.
NIULiubin, LIUJinzhao, XIAOBinghuan, et al. Study on the Rail Deflection Characteristics Excited by the Track Profile Irregularity of High Speed Railway Lines [J]. China Railway Science, 2022, 43 (2): 28-39. in Chinese
[32]
张洋.基于绝对里程校准的动检数据异常检测算法研究[D].北京:北京交通大学,2022.
[33]
ZHANGYang. Research on Anomaly Detection Algorithm of Dynamic Inspection Data Based on Absolute Mileage Calibration [D]. Beijing: Beijing Jiaotong University, 2022. in Chinese