王日尧,徐鹏,张晓阳,高彦嵩,康庆涛

• 1:北京交通大学先进轨道交通自主运行全国重点实验室
• 2:中国国家铁路集团有限公司工电部
• 3:中国国家铁路集团有限公司科技和信息化部

摘要(Abstract)：

检测轨道不平顺对掌握轨道质量状况和安排维修作业至关重要。高速铁路轨道不平顺主要依靠动检车检测，检测周期一般为10或15 d,检测频次较低且检测成本较高。为实现高频次、低成本检测，提出一种基于逆滤波器和卷积神经网络的轨道高低不平顺检测方法。该方法以列车运行过程中采集的垂向加速度、水平加速度和速度作为输入，首先利用逆滤波器描述列车响应与轨道不平顺之间的输入输出关系，随后通过卷积神经网络提取响应特征，并分别输出1.5～42、1.5～70、1.5～120 m三个波长范围的高低不平顺。利用中国四条干线铁路上的动检车实测数据对所提方法进行验证，试验数据涵盖不同运行速度以及有砟、无砟道床形式。结果表明，三个波长范围高低不平顺的平均绝对误差分别为0.186、0.231、0.333 mm,相关系数分别为0.88、0.91、0.93;检测误差两倍标准差的最大值为0.876 4 mm,满足高低检测技术指标要求(≤1.2 mm)。研究结果表明，该方法具有较好的工程适用性，可为轨道状态高频监测、质量评估及维修决策提供技术支持。

关键词(KeyWords)： 高速铁路;高低不平顺;车体响应;FIR滤波器;卷积神经网络

Abstract:

Keywords:

基金项目(Foundation): 中国国家铁路集团有限公司科技研究开发计划(N2024G054);; 中央高校基本科研业务费专项(2025JBXT003)

作者(Author): 王日尧,徐鹏,张晓阳,高彦嵩,康庆涛

参考文献(References)：

• [1] 马帅.基于车辆响应的轨道几何状态评价方法研究[D].北京：北京交通大学，2020.Ma Shuai.Research on assessment method of track geom-etry state based on vehicle response [D].Beijing:Beijing Jiaotong University,2020.
• [2] Hao X L,Yang J,Yang F,et al.Track geometry estimation from vehicle-body acceleration for high-speed railway using deep learning technique [J].Vehicle System Dynamics,2023,61(1):239-259.
• [3] 中国铁路总公司运输局工务部.铁路工务技术手册：道岔 [M].北京：中国铁道出版社，2017.Track Department,Transportation Bureau of China Railway Corporation.Technical manual for railway works:turnouts [M].Beijing:China Railway Publishing House,2017.
• [4] 曲铭，许玉德.基于小波的列车加速度和轨道不平顺关系分析 [J].华东交通大学学报，2008,25(5):36-41.Qu Ming,Xu Yude.An analysis of the relationship of acceleration and track irregularity based on wavelet [J].Journal of East China Jiaotong University,2008,25(5):36-41.
• [5] 杨怀志，牛留斌，谷永磊，等.基于空间状态辨识理论的高速铁路车体垂向加速度预测模型 [J].铁道建筑，2020,60(2):110-115.Yang Huaizhi,Niu Liubin,Gu Yonglei,et al.Prediction model of high speed railway car-body vertical acceleration based on space-state identification theory [J].Railway Engineering,2020,60(2):110-115.
• [6] 何庆，汪健辉，李晨钟，等.关联车体响应的轨道不平顺各项指标相对权重分析 [J].北京交通大学学报，2022,46(3):80-87.He Qing,Wang Jianhui,Li Chenzhong,et al.Relative weights analysis of track irregularity indexes associated with vehicle body response [J].Journal of Beijing Jiaotong University,2022,46(3):80-87.
• [7] 马帅，刘秀波，张彦博，等.基于车辆响应的高速铁路轨道几何状态评估方法 [J].铁道学报，2023,45(11):117-127.Ma Shuai,Liu Xiubo,Zhang Yanbo,et al.Track geometry state assessment method for high-speed railway based on vehicle response [J].Journal of the China Railway Society,2023,45(11):117-127.
• [8] 郝晓莉，樊志兴，杨飞，等.基于FEDformer的高速铁路轨道不平顺和车辆车体加速度关联模型 [J].铁道学报，2025,47(10):62-69.Hao Xiaoli,Fan Zhixing,Yang Fei,et al.Correlation model of high-speed railway track irregularities and vehicle-body vibration accelerations based on FEDformer [J].Journal of the China Railway Society,2025,47(10):62-69.
• [9] 彭楠，刘金朝，成棣，等.轮轨周期性激励与车体垂向加速度相关性等效模型 [J].铁道学报，2026,48(3):36-45.Peng Nan,Liu Jinchao,Cheng Di,et al.Equivalent model for correlation between periodic wheel-rail excitation and vehicle body vertical acceleration [J].Journal of the China Railway Society,2026,48(3):36-45.
• [10] 肖源杰，邓子璇，周思嘉，等.基于PDEM的高速列车-有砟轨道-路基耦合系统随机动力响应分析 [J].铁道学报，2026,48(4):40-55.Xiao Yuanjie,Deng Zixuan,Zhou Sijia,et al.Stochastic dynamic response analysis of high-speed train-ballasted track-subgrade coupled system based on PDEM [J].Journal of the China Railway Society,2026,48(4):40-55.
• [11] Lee J S,Choi S,Kim S S,et al.A mixed filtering approach for track condition monitoring using accelerometers on the axle box and bogie [J].IEEE Transactions on Instrumentation and Measurement,2012,61(3):749-758.
• [12] Odashima M,Azami S,Naganuma Y,et al.Track geometry estimation of a conventional railway from car-body acceleration measurement [J].Mechanical Engineering Journal,2017,4(1):16-498-16-00498.
• [13] Plesner A,Engsig-Karup A P,True H.Detecting railway track irregularities with data-driven uncertainty quantification [J].Highlights of Vehicles,2025,3(1):1-14.
• [14] Li C Z,He Q,Wang P.Estimation of railway track longitudinal irregularity using vehicle response with information compression and Bayesian deep learning [J].Computer-Aided Civil and Infrastructure Engineering,2022,37(10):1260-1276.
• [15] Zhang H C,Chin Z Y,Borghesani P,et al.Evaluation of onboard sensors for track geometry monitoring against conventional track recording measurements [J].Measurement,2024,229:114354.
• [16] Kraft S,Causse J,Coudert F.Vehicle response-based track geometry assessment using multi-body simulation [J].Vehicle System Dynamics,2018,56(2):190-220.
• [17] 铁工电[2023]106号高速铁路线路维修规则[S].Tiegongdian [2023] No.106.Rules for Maintenance of high-speed railway lines [S].
• [18] Zhai W M.Vehicle-track coupled dynamics models [M]//Vehicle-track coupled dynamics:theory and applications.Singapore:Springer,2020:17-149.
• [19] Le C Y,Bottou L,Bengio Y,et al.Gradient-based learning applied to document recognition [J].Proceedings of the IEEE,1998,86(11):2278-2324.
• [20] Zhao X,Han X,Su W J,et al.Time series prediction method based on Convolutional Autoencoder and LSTM [C]//2019 Chinese Automation Congress (CAC).Piscataway:IEEE,2019:5790-5793.
• [21] Tang W,Long G,Liu L,et al.Omni-scale cnns:a simple and effective kernel size configuration for time series classification [PP/OL].V3.arXiv (2022-06-17)[2025-12-19].https://arxiv.org/abs/2002.10061.
• [22] 牛留斌，刘金朝，王卫东.长波高低不平顺与车体垂向加速度关联模型研究 [J].铁道学报，2015,37(4):75-79.Niu Liubin,Liu Jinzhao,Wang Weidong.Research on relationship model between long-wave profile irregularity and car-body vertical acceleration [J].Journal of the China Railway Society,2015,37(4):75-79.
• [23] Gholamalinezhad H,Khosravi H.Pooling methods in deep neural networks,a review [J].arXiv (2020-09-16)[2025-12-19].https://arxiv.org/abs/2009.07485.
• [24] 刘伶萍，魏世斌，赵延峰，等.GJ-6型轨道检测系统的研制与验证 [J].铁路技术创新，2015(2):53-56.Liu Lingping,Wei Shibin,Zhao Yanfeng,et al.Development and verification of GJ-6 track detection system [J].Railway Technical Innovation,2015(2):53-56.