基于声发射的高铁齿轮箱金属材料疲劳损伤分析  

Fatigue damage analysis of high-speed railway gearbox metal materials based on acoustic emission

作　　者：艾轶博 耿梦影 吕涛 张卫冬 AI Yibo;GENG Mengying;LÜTao;ZHANG Weidong(National Center for Materials Service Safety,University of Science and Technology Beijing,Beijing 100083,China;Southern Marine Science and Engineering Guangdong Laboratory,Zhuhai 519080,China)

机构地区：[1]北京科技大学国家材料服役安全科学中心,北京100083 [2]南方海洋科学与工程广东实验室(珠海),广东珠海519080

出　　处：《铁道科学与工程学报》2023年第2期423-431,共9页Journal of Railway Science and Engineering

基　　金：南方海洋科学与工程广东省实验室(珠海)创新团队项目(311021013);中央高校基本科研业务费基金资助项目(FRF-GF-20-24B,FRFMP-19-014)。

摘　　要：声发射作为一种现代无损检测技术,广泛应用于金属材料或设备的实时状态监测,对金属材料或设备的服役安全评价具有重要意义。作为高铁的重要组成部分,齿轮箱体的服役情况直接影响列车的安全性。针对高速列车齿轮箱的故障诊断与安全预警问题,传统的方法对齿轮箱零部件的早期失效不敏感,故障准确定位困难。因此,提出一种基于声发射技术对齿轮箱金属材料服役状态实时监测的方法。搭建带声发射检测功能的箱体材料疲劳试验系统,通过疲劳试验数据计算得到安全阶段与预警阶段的临界时间,对齿轮箱体材料疲劳实验过程中直至完全断裂时刻的声发射信号进行采集与分析,提取出能量与振铃计数的比值作为反映材料当前退化状态的主要特征量。分别采用支持向量机(SVM)和加权支持向量机(WSVM)构建分类器,并利用粒子群优化算法(PSO)进行参数寻优,对齿轮箱箱体材料疲劳裂纹扩展的各阶段进行损伤分类。实验结果表明,在齿轮箱箱体金属材料失效数据匮乏的情况下,SVM的分类结果严重偏向于安全阶段,而WSVM很好地兼顾了安全阶段与预警阶段的识别分类,对材料当前所处损伤状态分类的平均准确率超过90%,可以对金属材料的失效做出预警,为今后的在线监测奠定基础。As a modern non-destructive detection technology, acoustic emission has been widely used in real-time state monitoring of metal materials or equipment, and has great significance for the in-service safety evaluation of metal materials or equipment. The service condition of gearbox body, as an important component of high-speed trains, directly affects the safety of trains. According to the fault diagnosis and early warning problems of the high-speed train gearbox, the traditional method is insensitive to the early failure of the gearbox parts, and the accurate fault positioning is difficult. Therefore, we proposed a method for real-time monitoring of gearbox materials using acoustic emission technology to solve the problem of real-time early warning of critical components for high-speed trains. A gearbox material fatigue test system with acoustic emission monitoring system was built. The critical time of the safety and early warning stages were calculated from the fatigue test data, and acoustic emission signalscollected during the material fatigue experiments up to the moment of complete fracture were analyzed to extract the ratio of energy to ringing count as the maincharacteristic quantities that can reflect the current degradation state of the gearbox material. The classifier was constructed using support vector machine(SVM) and weighted support vector machine(WSVM), respectively, and a particle swarm optimization(PSO) algorithm was utilized for parameter optimization to classify the damage identification at each stage of the fatigue crack expansion process in gearbox materials. The experimental results show that the classification results of SVM are heavily biased towards the safety phase in the absence of data on the failure of the metal materials of the gearbox case. The WSVM took into consideration the identification and classification of the safety and warning phases well, and the average accuracy of classification of the current damage state of the material exceeded 90%. The findings can provide early wa

关 键 词：声发射 疲劳损伤 支撑向量机 模式识别 安全预警 

分 类 号：U214.8[交通运输工程—道路与铁道工程]