外文原文-脉冲涡流检测中的频域缺陷特征.pdf

1、International Journal of Applied Electromagnetics and Mechanics 45(2014)621625621DOI 10.3233/JAE-141885IOS PressFrequency-domain defect characterization inpulsed eddy current testingZhiwei Zeng,Yansong Li,Lin Huang and Minfang LuoDepartment of Aeronautics,Xiamen University,Xiamen,Fujian,ChinaAbstrac

2、t.Pulsed eddy current(PEC)testing has attracted researchers interest because the pulsed excitation comprises a broadband of frequencies and the response signal provides more information about defect than traditional eddy current testing.Various features have been extracted from PEC signal for defect

3、 characterization.In this paper,we extract frequency-domainfeatures and propose defect characterization scheme for identifying defects location,radius,and height.Keywords:Pulsed eddy current testing,defect characterization,feature1.IntroductionIn pulsed eddy current(PEC)testing,the coil is excited b

4、y pulsed waveform,typically rectangularwave,as illustrated in Fig.1.PECtesting hasmanyadvantagesovertraditional eddycurrent(EC)testing,such as better field penetration and less sensitive to liffoff 1.More importantly,induced currents in thetest sample have a wide band of frequency components.Due to

5、the skin effect,different frequencycomponents have different penetration depths.Therefore the response signal in PEC testing providesricher depth information about defect than that of traditional EC testing.As a result,there are moreoptions for defect characterization in PEC testing.In the recent ye

6、ars,studies of PEC testing have been focused to the extraction of features from thetransient response signal,which is critical to defect characterization.He et al.presented time-domainanalysis method,which extracts time-domain features of transient signal after difference processing,asshown in Fig.2

7、 2.Tian et al.developed wavelet-based principal component analysis(PCA)method,which applies PCA to the wavelet coefficients of PEC signal to extract dominant features 3.Some re-searchers transformed PEC signals to the frequency domain and performed spectral analysis.Vasi c et al.analyzed signal spec

8、tra but did not use frequency-domain features for defect classification 4.Lebrunet al.used a particular frequency,namely characteristic frequency,as a feature to characterize defectlength 5.Yang et puted energy of signal in the low-frequency band and used it as a feature 6.Both 5,6 combined frequenc

9、y-domain features with time-domain features for defect characterization.Basedon the fact that different frequencycomponentshavedifferent penetrationdepths,it is advisabletostudy PEC defect characterization with all the features being spectral features,which has more physicalCorresponding author:Zhiw

10、ei Zeng,Department of Aeronautics,Xiamen University,Xiamen 361005,Fujian,China.E-mail:1383-5416/14/$27.50 c?2014 IOS Press and the authors.All rights reserved622Z.Zeng et al./Frequency-domain defect characterization in pulsed eddy current testingrhMagnetic field sensorFig.1.Pulsed eddy current testi

11、ng.O tPeak valueZero-crossing time Differential output Fig.2.Typical PEC transient signal and time-domain features.significance than the other characterization schemes.He et al.transformed transient signal to the fre-quency domain and select high-frequency components as features 1,7.However,this cha

12、racterizationscheme will lose deeply hidden defects which are detectable only by low-frequency components.In this paper,we attempt to fully exploit frequency-domain information of PEC signal for defect char-acterization.More specifically,not only high-frequency component,but also low-frequency compo

13、nentand middle-frequency information are extracted.Simulation results show that the frequency-domainfeatures are promising to identify defects location,size,and depth.The idea is different from multi-frequency eddy current(MEC)testing.In MEC testing 810,optimization of frequencies in terms ofquantif

14、ying defect depth is impossible as defect depth is not known a prior in real situation.On thecontrary,appropriate frequency components of PEC signal can be selected from the spectrum to bestcharacterize defect parameters.2.Finite element analysisThe signals used in the paper are generated by numeric

15、al simulation using the finite element method.The Fourier transform method is utilized for obtaining transient signal and its spectrum.2.1.Geometry and parametersIn Fig.1,the air-core coil has inner radius of 6 mm,outer radius of 10 mm,height of 13 mm,andliftoff of 1 mm.The coil has 312 turns and is

16、 excited by rectangular wave of 100 Hz,50%duty ratio,and0.333 A amplitude.The sample is 4 mm thick.The conductivity and relative permeability of the sampleare 58.6%IACS and 1,respectively.Defect(assumed cylindrical type in the paper)is characterized by three parameters:location(top orbottom,or surface or subsurfacein other words),radiusr,and heighth.We use notationTrh(Brh)to denote top(bottom)defect with radiusrand heighth.The unit ofrandhis mm.2.2.Fourier transform methodTransient modeling usin