3D Shape Identification of Perpendicular Flaw by Biazial MFLT with Neural Network

摘要：

In this paper, we attempt to estimate the dimension of a parallelepiped flaw and to identify the location, I. E., the horizontal position and the located surface, by biaxial MFLT with a neural network. The specimen is a magnetic material subjected to the magnetic field, and the magnetic flux in the specimen leaks in the vicinity of the flaw. We measure the biaxial MFL, i.e., the tangential and the normal components of MFL field along a line parallel to the specimen surface. Then we approximate the measured biaxial MFL distributions near the flaw by elementary functions with a few parameters, say, MFL parameters. The horizontal position of a flaw along the measured line is characterized by some of MFL parameters. Neural network is used to infer the cross section of the flaw, i.e., the width, the depth and the located surface from the MFL parameters. By repeating a similar process along several lines parallel to the specimen surface, we can identify the three dimensional shape of the flaw including its horizontal position. The neural network trained with respect to several known flaws is found to evaluate the three dimensional shape and location of a flaw in a good accuracy.

关键词： NDT MFLT Neural network Flaw detection Magnetic impedance sensor

作者: Abe Masataka Biwa Shiro Matsumoto Eiji

作者单位: Department of energy conversion science, Kyoto university, Kyoto 606-8501, Japan

会议类型: 国际会议

会议名称: ISSNP2008、CSEPC、ISOFIC2008(第二届21世纪和谐核电系统国际会议、第四届电厂控制中认知系统工程方法国际会议暨第三届未来核电厂仪表与控制国际会议)

会议地点: 哈尔滨

会议语种:英文

页码: 106-112

在线出版日期: 2008-09-08（万方平台首次上网日期，不代表论文的发表时间）