Study On Quantitative Method Of Subsurface Defect Depth For Conductor Structure

Conductive structural parts are widely used in aerospace,rail transportation,military defense and other fields.Timely and reliable detection of conductive structural parts is of great significance to ensure the safe operation of equipment and avoid the occurrence of major accidents.Pulsed eddy current and ultrasonic testing are two mature nondestructive testing methods.The former has strong ability to detect the defects on the surface of conductive structure and on the surface However,ultrasonic detection has a blind area,which is more difficult to detect surface and subsurface defects,and more suitable for deeper defects detection.It can be seen that the two methods complement each other in detection depth,but there are limitations in the detection of subsurface defects,and there is no single detection method that can solve the problem of subsurface defects detection,in particular,the quantitative problem of the depth of subsurface defects.In this paper,we focus on the quantitative analysis of subsurface defect depth of conductive structure,and study the method of pulsed eddy current and ultrasonic testing:1)For PEC/UT detection signal contains complex mixed noise,the signal de-noising method,this paper puts forward a kind of based on ABC-VMD de-noising algorithm,the traditional VMD algorithm is improved based on the recursive energy variation mode decomposition algorithm,and then artificial colony algorithm based on energy VMD layers and punishment factor combination optimization,to obtain the optimal modal component,to reconstruct the signal.2)To solve the problem that the PEC/UT defect detection ability is different at different depths,the adaptive feature weight allocation method is studied,and a fusion algorithm of feature weight based on statistical distribution and distance measure is proposed.By constructing the dispersion model and distance measure model,the statistical distribution weight and distance measure weight of features are calculated.Finally,D-S evidence theory is used to combine the weights of the two different algorithms to obtain the best feature weights3)For single detection method cannot satisfy the conductive structure of surface defect detection problem,the PEC/UT data fusion method,this paper proposes a multiple criteria weights based on D-S evidence fusion method,through the calculation of information entropy,the conflict of evidence distribution of entropy and entropy build credibility evaluation model,a reasonable weight distribution of the evidence,again to weight optimization of redundant probability redistribution,the final obtained from D-S weighted fusion joint probability distribution and the quantitative results.