| The ferromagnetic materials(e.g.the carbon steel)manufactured equipments such as chemical tank,energy transportation pipeline,large-scale mechanical equipment will emerge corrosion defects during long term running.The damage accumulation will lead to the failure of equipments and unexpected shutdown,which cause economic property lose and even casualty.Hence,it is particularly important to conduct regular nondestructive testing for the key parts of these devices.Pulsed eddy current testing technology has been widely applied in deep defect detection of metal components due to its characteristics,such as large lift-off,non-contacting,easy signal interpretation and so on.In practical applications,the method of reducing the pulse excitation frequency is adopted to increase the detection depth,however,it will cause the problems of weak signal which lead to difficult signal collecting and processing in the middle and late stages.In this thesis,on purpose of the detection system design,signal de-noising method and feature quantity extraction for the characteristics of weak signal and large background noise in the middle and late stage of pulsed eddy current testing of ferromagnetic materials,the main research work are as follow:(1)The design of pulsed eddy current testing system.A pulsed eddy current testing platform based on virtual instrument was constructed.The hardware system based on NI data acquisition card and the upper computer software system based on LabVIEW were designed.The hardware significantly concentrated on the probe module and signal conditioning amplification module to improve the acquisition and anti-interference ability of the weak signals.The software system implemented capabilities including generation of excitation data acquisition and control signal display and recording,etc.are realized.The system integrated excitation generation and acquisition control,which can meet the high-efficiency requirement of the pulsed eddy current signals under large-dynamic range.(2)The research on de-noising method of pulsed eddy current testing signal.Pulsed eddy current(PEC)testing signals are often interfered by strong background noises,which make it difficult to efficiently extract signal features.A novel PEC signal de-noising method is proposed based on the Wiener filtering and principal component analysis(PCA).The method combined the advantages of Wiener filtering and principal component analysis for signal de-noising.Firstly,the weak signal under strong noise is selected with suitable template window for Wiener filtering preprocessing,the signal noise ratio is improved while maintain low distortion.Then,the preprocessed signals are subtracted by a reference signal to eliminate the system noise.Afterwards,the principal components of differential signals are extracted byusing PCA,Afterwards,the principal component features of differential signals are extracted by using PCA,and adequate number of principle components at a given threshold is selected to perform the signal reconstruction scheme.Following the above de-noising procedures,a time-domain signal with high signal-to-noise ratio(SNR)is finally obtained.This method is used to analyze the testing signal of Q235 step board,and the results show that strong noises in experimental signals are effectively eliminated and the SNR is significantly improved to95.5 dB.(3)The extraction and quantitative evaluation of the defect signal features.The designed pulsed eddy current testing system was applied on three kinds of test pieces with different depth square grooves,different diameter round-hole and different thickness steps.The detection signal was processed and analyzed from both the time-domain and the time-frequency domain,and the feature quantities under different analysis domains were extracted.In the time-domain,the differential peak of the detection signal can be used to quantitatively evaluate the size of the three types of defects.In the time-frequency domain,after the signal was smoothed by the pseudo Wigner-Ville distribution,the diffusion of the low-frequency component along the time axis is related to the defect size.In practically detection,especially under the strong noise background,the comprehensive application of the time domain and time-frequency domain feature quantity can effectively improve the accuracy of defect evaluation. |
PDF Full Text Request