The Application Research of Coating Thickness Detection Using The Tree Classifier

摘要：

　　Using Ultrasonic Testing to test the coating thickness of various industrial structure is a commonly used thickness measurement method.But in some special structures,such as steel in the solid rocket motors,insulation structure,the coating thickness could only be tested by the Ultrasonic emitted from the outside of the steel,This paper presents a method of using resonant relationship of the two detection echo as the delay time of eigenvalue,and make the eigenvalue extraction boil down to a multi-pattern recognition problem,using a self-design binary tree classifier to classify them,effectively improving the feature extraction accuracy and the signal to noise ratio of the thickness detection.Using Ultrasonic testing to test the coating thickness of various industrial structure is a commonly used thickness measurement method,but in some special structures,such as steel in the solid rocket motors.Insulation structure,the coating thickness could only be tested by the Ultrasonic emitted from the outside of the steel.The time domain aliasing of the ultrasonic detection signals happens when the Coating thickness is small,it results in the extraction of the thickness eigenvalue difficult,even could not be extracted.On the other hand,similar to the inside bonding of rubber coating of the steel,there is a large mismatch between the high acoustic reactance of the metal and the low acoustic reactance of the rubber,and because of the high sound attenuation of the rubber-base material,it leads to weakness of the detection signals and the difficult to thickness eigenvalue extraction.this paper presents a method of using oblique ultrasonic pulse echo in immersion to test the thickness of the rubber layer in the steel-rubber coating structure and the resonant relationship of the two detection echo as the delay time of eigenvalue,and make the eigenvalue extraction boil down to a multi-pattern recognition problem,using a self-design binary tree classifier to classify them,effectively improving the feature extraction accuracy and the signal to noise ratio of the thickness detection.

关键词： Ultrasonic testing Thickness Amplitude Spectrum Tree classifier

作者: J.T.Zhang X.G.Shen

作者单位: College of Mechanical Engineering & Automation, The North University of China, Taiyuan, China,030051

会议类型: 国际会议

会议名称: the 2011 International Conference on Frontiers of Manufacturing Science and Measuring Technology (第一届制造科学与检测技术国际会议(ICFMM2011))

会议地点: 重庆

会议语种:英文

页码: 1034-1038

在线出版日期: 2011-06-23（万方平台首次上网日期，不代表论文的发表时间）