摘要
针对高速列车运营维护中的车轮多边形在线监测与磨耗识别问题,提出一种基于声信号的变分经验模态分解(VMD)结合粒子群优化支持向量机(PSO-SVM)车轮磨耗识别方法。首先,对静态时车轮多边形磨耗水平进行测试,并采集高速列车平稳运行时车内噪声数据。其次,分析车内噪声与车轮多边形磨耗幅值的数据规律,将车内噪声与车轮多边形的关系进行映射。随后,应用PSO算法搜寻VMD最优分解参数,结合带通滤波滤除冗余的噪声频段,提取时域和频域特征指标。最后,应用PSO算法优化SVM最优模型参数组合,实现将VMD算法的信号分解能力和支持向量机识别能力的有效结合。实验验证结果表明,基于声信号的VMD结合PSO-SVM车轮磨耗识别方法能有效根据车内噪声信号识别转向架车轮最大磨耗幅值,为动车车轮镟修提供指导和帮助。
On-line monitoring and identification of wheel polygon wear is one of the important problems to be solved in high-speed train operation and maintenance.A novel wheel wear identification method combined variational empirical mode decomposition(VMD)and particle swarm optimization support vector machine(PSO-SVM)based on sound signals is proposed in this paper.Firstly,the static wheel polygon wear level is tested and the in-vehicle noise data of high-speed train is collected.Secondly,the data rules of interior noise and wheel polygon wear amplitude are analyzed,and the relationship between interior noise and wheel polygon is mapped.Thirdly,the PSO algorithm is applied to search the optimal decomposition parameters of VMD,and the redundant noise frequency band is filtered by band pass filtering.Then the time domain and frequency domain feature indexes are extracted.Finally,the PSO algorithm is used to optimize the optimal model parameter combination of SVM,and the signal decomposition capability of VMD algorithm is effectively combined with the recognition capability of support vector machine.The experimental verification results showed that the proposed wheel wear identification method could effectively identify the maximum wear amplitude of bogie wheels according to the noise signal inside the vehicle.The study provides guidance and help for wheel rotation and repair of high speed train.
作者
冯前前
刘兴起
李鹏震
许海荣
韩春刚
石邹亮
刘运航
FENG Qianqian;LIU Xingqi;LI Pengzhen;XU Hairong;HAN Chungang;SHI Zouliang;LIU Yunhang(China Railway Urumqi Group Co.,Ltd.,Urumqi 830000,China;State Key Laboratory of Rail Transit Vehicle System,Southwest Jiaotong University,Chengdu 610031,China)
出处
《机械》
2025年第6期31-39,共9页
Machinery
基金
中国国家铁路集团有限公司科技研究开发计划(N2023J036)。
关键词
车轮多边形
支持向量机
粒子群算法
变分经验模态分解
wheel polygon wear
support vector machine
particle swarm optimization algorithm
variational empirical mode decomposition
