Identification And Accurate Location Of Hidden Damage For Welded Joints Based On Metal Magnetic Memory Technology

Welded joints are common structures,which play an important role in the safe operation of equipment.As a new nondestructive testing technology,metal magnetic memory(MMM)testing can not only be applied to the macroscopic damage detection of the magnetic materials,but also can detect the stress concentration and the early hidden damage.However,due to the uncertainty and dispersion of MMM signal,there are still some problems to be solved.This paper starts from the mechanism of MMM,the fatigue experiment is first operated.The experimental materials are prefabricated incomplete fusion,lack of penetration and porosity.The MMM feature is observed in the process of damage evolution by MMM curve and space figure,which provides the experimental basis for quantitative damage identification and location.In order to quantitatively identify hidden damage state of welded joints by using metal magnetic memory(MMM)technology,the modified maximum likelihood estimation(MLE)MMM model is first proposed.The experimental materials are Q235 B welded plate specimens.The MMM signals are scanned by the TSC-2M-8 in the fatigue tension experiments with synchronous X-ray detection.It has been found that the MMM signals appear uncertainty and it's difficult to distinguish different damage states only by a single MMM parameter.So the six MMM characteristic parameters,?Hp(x),Kxmax,mxmax,?Hp(y),Kymax and mymax,are extracted corresponding to the three damage states,that is,the crack free,the hidden crack and the macroscopic crack,respectively.The probability values of the above six parameters are calculated by the optimized bandwidth kernel density estimation.Then the MMM identifying model of the three damage states is established based on the MLE method.Furthermore,considering that the identified results of the MLE MMM model exist partial overlaps,the modified MLE MMM model is presented on the basis of D-S theory.The verification result shows the uncertainty is as low as 0.3%,which provides a new tool for identifying damage states of welded joints quantitatively.In order to locate accurately the hidden damage of welded joints by using metal magnetic memory(MMM),the MMM gradient model is present based on maximum likelihood estimation(MLE)optimized by particle swarm optimization(PSO).The fatigue tensile experiments were operated to find the relation between MMM gradient and distance from defect of welded joints by comparing to the X ray detection.The experiment material is Steal Q235 prefabricated with incomplete penetration.Then the nonlinear function is built byposition parameter and MMM gradient.The MLE is introduced to establish the objective function considering the locating result of the hidden damage.Furthermore,the nonlinear objective function is easy to get into local extremum instead of global extremum.So the PSO is adopted to optimize the objective function for the global search ability.The validation results of the model present the locating error is 3.48%,which provide a new idea for the practical application of MMM technology to detect early hidden damage and locate accurately.