摘要
脉冲涡流检测广泛应用于工业设备的无损检测中,针对当前脉冲涡流无损检测中仍存在缺陷定量化难的问题,采用新一代磁阻传感器-隧道磁阻传感器设计阵列化脉冲涡流检测探头,对圆形缺陷开展了检测实验,并采用BP神经网络算法定量化评估圆形缺陷,最后验证了该定量化方法对非圆形缺陷的适用性。研究表明阵列化脉冲涡流检测探头差分信号峰值服从高斯分布,且高斯分布标准差与缺陷宽度之间以及差分信号峰值时间、上升时间与缺陷深度之间都存在一定的映射关系,因此可建立BP神经网络估计缺陷宽度和深度,且估计精度较高,这为更复杂缺陷的定量化检测研究提供了一定的参考。
Pulse eddy current testing is widely utilized in the nondestructive testing of industrial equipment.To address the problem of quantifying defects in the nondestructive testing of pulse eddy current,a new generation of magnetoresistive sensor and a tunnel magnetoresistive sensor are used to design the array probe.The testing experiment on circular defects is conducted.Then,the quantification of circular defects is performed by using back propagation neural networks and the applicability of this quantifying method on non-circular defects is carried out.Research results indicate that the peak values of differential signals acquired by the probe obey Gaussian distribution.The mapping functions between standard deviation of Gaussian distribution and widths of defects,as well as among peak time,rise time of differential signal and depths of defects are identified.Therefore,the back propagation neural networks can be built to estimate widths and depths of defects and the accuracy of the estimation is high,which provides a reference for quantitative research of more sophisticated defects.
作者
梁远远
杨生胜
文轩
银鸿
黄乐程
Liang Yuanyuan;Yang Shengsheng;Wen Xuan;Yin Hong;Huang Lecheng(Science and Technology on Vacuum Technology and Physics Laboratory,Lanzhou Institute of Physics,Lanzhou 730000,China)
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2018年第11期70-78,共9页
Chinese Journal of Scientific Instrument
基金
“十三五”全军共用信息系统装备预先研究课题(31512030101)项目资助
关键词
脉冲涡流无损检测
隧道磁阻
BP神经网络
缺陷定量化
pulse eddy current nondestructive testing
tunneling magnetoresistance
back propagation neural networks
quantification of defect
作者简介
梁远远,2016年于北京航空航天大学获得学士学位,现为兰州空间技术物理研究所硕士研究生,主要研究方向为涡流无损检测。E-mail:lyy12151119@163.com;通信作者:杨生胜,2003年于中国空间技术研究院获得博士学位,现为兰州空间技术物理研究所研究员、博士生导师,主要研究方向为弱磁测量、涡流无损检测和空间环境效应及防护技术研究。E-mail:2syang@sina.com
