摘要
地铁钢轨短波波磨现象严重影响列车运行安全,更快速、准确地对钢轨波磨进行检测,有利于及时指导钢轨打磨,从而避免或减少由钢轨波磨引发的一系列问题。文章以轮轨噪声作为检测信号,提出了一种基于支持向量机(SVM)的地铁钢轨短波波磨特征识别框架;结合轮轨噪声和短波波磨类别特点,采用时域-频域特征提取方法,以最大化支持向量机分类精度为依据,实现对特征的有效提取和选择;较为全面地考虑现实中的各类钢轨短波波磨类型,实现对短波波磨的正确分类。分类测试结果表明,基于轮轨噪声和支持向量机的地铁钢轨短波波磨特征识别方法能够有效地对波磨波长和幅值进行正确分类,其中波长分类平均精度达到97.32%,幅值分类平均精度达到97.99%.
The phenomenon of short-pitch corrugation for metro rail seriously affects the safety of train operation. The faster and more accurate detection of rail corrugation is helpful for guiding rail grinding in time,so as to avoid or reduce a series of problems caused by rail corrugation. Taking wheel-rail noise as the detection singal, this paper proposes a feature recognition framework of short-pitch corrugation for metro rail based on support vector machine(SVM). Combining the characteristics of wheel-rail noise and short-pitch corrugation, a time-frequency domain feature extraction method is adopted to achieve effective feature extraction and selection of features based on maximizing the classification accuracy of support vector machine. More comprehensive consideration of the reality of all types of rail short-pitch corrugation, to achieve the correct classification of shortpitch corrugation. The classification test results show that the feature recognition method of short-pitch corrugation for metro rail based on wheel-rail noise and support vector machine can effectively classify the wavelength and amplitude of corrugation correctly, and the average accuracy of wavelength classification is 97.32%, and the average accuracy of amplitude classification is 97.99%.
作者
刘晓龙
温泽峰
肖新标
陶功权
谢清林
LIU Xiaolong;WEN Zefeng;XIAO Xinbiao;TAO Gongquan;XIE Qinglin(State Key Laboratory of Traction Power,Southwest Jiaotong University,Chengdu 610031,China)
出处
《电力机车与城轨车辆》
2023年第1期36-42,共7页
Electric Locomotives & Mass Transit Vehicles
基金
城市轨道交通数字化建设与测评技术国家工程实验室开放课题基金(2021JZ04)。
关键词
钢轨波磨
轮轨噪声
支持向量机
特征提取
rail corrugation
wheel-rail noise
support vector machine
feature extraction
