摘要
针对轨道几何参数检测数据为空间域等间隔采样序列的特性,将快速动态时间规整的方法引入到空间域中,研究单边轨距与曲率之间数据变化趋势的相似性。首先利用S⁃G滤波器对单边轨距进行数据平滑,根据最佳线性相关度选取滤波器的滑动窗口长度和多项式拟合阶次;然后利用快速动态时间规整方法计算归一化数据平滑后的单边轨距与归一化曲率的动态规整距离,实现两者之间的相似性度量。结果表明,健康样本与故障样本中的单边轨距与曲率的动态规整距离表现出了较大的差异性,该相似性研究在故障判别中具有较好的应用潜力。
In view of the characteristics that the track geometry parameter detection data are equal-space sampling sequences in spatial domain,the method of fast dynamic time warping was introduced into spatial domain to study the similarity of data change trend between unilateral gauge and curvature.Firstly,S-G filter was used to smooth the unilateral gauge data,and the sliding window length and polynomial fitting order of the filter were selected according to the best linear correlation.Then,the fast dynamic time warping method was used to calculate the dynamic warping distance between the unilateral gauge after smoothing of normalized data and the normalized curvature,so as to realize the similarity measurement between them.The results show that there is a great difference between the dynamic warping distance of unilateral gauge and curvature in the healthy samples and the fault samples,and the similarity research has a good application potential in fault discrimination.
作者
郑祥爱
刘文波
蔡超
徐梦莹
王平
姚楠
ZHENG Xiangai;LIU Wenbo;CAI Chao;XU Mengying;WANG Ping;YAO Nan(College of Automation Engineering,Nanjing University of Aeronautics and Astronautics,Nanjing 211100,China;Non-Destructive Testing and Monitoring Technology for High-Speed Transport Facilities Key Laboratory of Ministry of Industry and Information Technology,Nanjing 211100,China;Infrastructure Inspection Research Institute,China Academy of Railway Sciences Corporation Limited,Beijing 100081,China)
出处
《铁道建筑》
北大核心
2021年第3期138-142,共5页
Railway Engineering
基金
国家重点研发计划(2018YFB2003304)
国家自然科学基金(61871218)
南京航空航天大学研究生创新基地(实验室)开放基金(kfjj20190306)
中央高校基本科研业务费专项资金(NJ2019007,NJ2020014)。
关键词
高速铁路
相似性
数值模拟
轨道几何参数
快速动态时间规整
空间域等间隔采样
high speed railway
similarity
numerical simulation
track geometry parameters
fast dynamic time warping
equal-space sampling in spatial domain
