动车组空气弹簧橡胶脱粘检测系统设计与研究  

Design and research on rubber debonding detection system of EMU air spring

作　　者：甘文成 王雪梅[1] 张永敬 侯大伟 Gan Wencheng;Wang Xuemei;Zhang Yongjing;Hou Dawei(School of Mechanical Engineering,Southwest Jiaotong University,Chengdu 610031,China;Shanghai Train Set Depot,China Railway Shanghai Group Co.,Ltd.,Shanghai 201812,China)

机构地区：[1]西南交通大学机械工程学院,成都610031 [2]中国铁路上海局集团有限公司上海动车段,上海201812

出　　处：《现代制造工程》2019年第10期129-134,共6页Modern Manufacturing Engineering

基　　金：上海铁路局科研计划项目(2018067).

摘　　要：为了实现动车组空气弹簧橡胶脱粘检测,建立了将超声脉冲回波检测方法与模式识别技术相结合的检测系统。传统的检测识别方法仅利用超声脉冲高次回波声压比,在此基础上,进一步分析提取了超声回波归一化幅频曲线的分布特征,使时域特征和频域特征的结合能够更全面地反映出金属-橡胶界面的粘接情况。通过对金属-橡胶试块和动车组空气弹簧样件进行实际超声C-扫描检测,并采用BP神经网络对提取的特征量进行训练与识别,结果显示,其对脱粘缺陷的识别率达到95%,表明所建立的检测系统能够准确地识别出金属-橡胶单层粘接界面的脱粘缺陷,能够满足动车组空气弹簧的检修要求。In order to realize the detection of rubber debonding of Electrical Multiple Units (EMU) air spring,a detection system combining ultrasonic pulse echo and pattern recognition technology was established.On the basis of traditional detection and recognition only using the sound pressure ratio of high order echo,the distribution characteristics of normalized amplitude-frequency curve of ultrasonic echo were further analyzed and extracted.The combination of time-domain features and frequency-domain features can reflect the bonding situation of metal-rubber interface more comprehensively.Through the actual ultrasonic C-scan detection of metal-rubber and EMU air spring samples,and using BP neural network to train and identify the extracted features,the recognition rate of debonding defects reach to 95 %.The results show that the system can accurately identify the debonding defects of the bonding interface of single metal-rubber layer,and can meet the maintenance requirements of EMU air spring.

关 键 词：空气弹簧 脱粘检测 特征提取 BP神经网络 超声C-扫描检测 

分 类 号：U279.335[机械工程—车辆工程]