The Research Of Crack Source’s Acoustic Emission Radiation And Classification

In today’s society, the problem of safety arouses people’s more and more attention in production and life, and the crack problem of the materials is the great hidden danger not to be ignored. The nondestructive testing for crack by using acoustic emission technology is an important research direction in modern research. In this field, the theoretical research of crack source’s acoustic emission radiation characteristics and classification of crack types is very important.In elastodynamics, Akiiti and others put forward the elastodynamics equation of the particle’s motion. On the basis of the equation P.Fellinger put forward the finite integration technique(EFIT) to solve the elastodynamics eqution from the point of numerical method. When using EFIT method, the data is computationally and the size is very large to meet the requirement of accuracy. The Open CL parallel technique can solve this problem effectively. The paper introduces the basic principle of Open CL and basic ideas applying Open CL to EFIT and contrast of execution efficiency in details.The paper introduces the moment tensor which describes the crack source ’s characteristic from the point of mechanics according to the elastodynamics equation of the particle’s motion, and deduces the model of the crack source ’s radiation field, and divides the model into near-field and intermediate-field and far-field. The paper researches radiation laws of the elastic waves produced by crack source from the point of the directivity. The research combines displacement field simulated by EFIT according to Open CL.For classification of the crack source types, the paper researches from the two main directions. According to the radiation characteristics of different crack source, we can set the locations of sensors to get the acoustic emission signal ’s first amplitude and make the feature vector, and then build the feature library of the typical cracks. To classify the cracks we put the feature vector of the detecting crack’s first amplitude and the vector in feature library in similarity test. The other method is building the inversion method on the basis of acoustic emission signal from the model of crack source’s radiation and get the moment-density-direction tensor for classifying crack sources. And the paper make the experimental verification for these two method.