Study On Detection Methods Aiming At Rail Tread Oblique Crack Defects Using Pulsed Eddy Current

The timely and effective nondestructive testing is one of important methods to guarantee the safety and integrity of key structures in the aerospace,rail transportation,industrial production,and many other fields.The thesis is aiming at the tread oblique crack detection difficulties.Under the support of the National Natural Science Foundation of China and practical engineering applications,the study on key technologies of pulsed eddy current testing in signal preprocessing,feature extraction,defect qualitatively and quantitatively evaluation,design and development of the pulsed eddy current(PEC)prototype system are carried out.The main research work and innovations are as follows:(1)The physical model,detection mechanism and response characteristics of pulsed eddy current testing are studied.The rail defect structure characteristics are analyzed.The typical PEC signals and ferromagnetic material PEC signals are theoretically derived and analyzed,which provide theoretical guidance to the defect detection and analysis.(2)After the analysis of PEC signal characteristics in the propagation angle and extend depth of the oblique crack defect,we see that PEC signals are affected by many external disturbances,and response characteristics are not obvious in time domain,and defects are not easy to inspect and discriminate just by simple time domain methods.For avoiding the above problems and making full use of frequency domain information,the PEC signal features of curvilinear integral in time domain,max peak value and frequency in the small peak value in Hilbert-Huang time-frequency domain are studied.Meanwhile,various features in time domain,frequency domain and time-frequency domain are extracted and analyzed.The information gain method is introduced to estimate the extracted features significance,and the extracted features are verified to be valid,which indicate that the amount of information obtained is greatly improved and lay foundation for the subsequent rail defect detection and quantitative evaluation work.(3)The qualitative and quantitative evaluation methods on rail tread oblique crack are studied.On the problem of the nonlinear complex relationship between the PEC signal features and defect type and parameters,the supported vector machine(SVM)method based on Radial Basis Function(RBF)is studied,which is applied to the qualitative analysis of the oblique crack extend angles.On the basis of the SVM analysis,the confidence probability in the SVM analysis process is used as the evidence,which is applied to the quantitative analysis of the oblique crack depth with the study of D-S evidence theory in multicycle fusion.Under the testing of the rail oblique crack,the problem of features overlap and nonlinearity in PEC signals can be solved by the combination of SVM and D-S evidence theory,which can be used to determine the propagation angle and extend depth of the oblique crack.(4)Optimization work is carried out based on the PEC prototype system in the laboratory.With the purpose of improving the detection sensitivity,electromagnetic field analysis based on Comsol software is carried out from the theoretical relationship between PEC response signals and probe size parameters such as height,radius and number of turns.Based on that,the testing probe size parameters are proposed for applying to the ferromagnetic material defect detection.Nondestructive testing application work is introduced using the improved PEC testing prototype system.In short,the research work in this dissertation may provide technical supports to the rail tread oblique cracking analysis as well as conductive structure nondestructive testing and evaluation in various fields.