Prediction Model Of Early Hidden Damage For Welded Joints Based On Metal Magnetic Memory Technology

Welding technology plays an irreplaceable role in the national economy and production. Due to the limitation of many factors such as the external environment during welding process, welding technical level, welding technology, there always exist typical defects in the welded joints. The types of the typical defects are slag inclusion, incomplete penetration. With the accumulation of the operation time during service, these defects will gradually evolved into damage defects and cause serious consequences. The conventional nondestructive testing techniques can only test for macroscopic defects. The metal magnetic memory(MMM) technology proposed in the1990 s can not only detect macroscopic defects, but also detect the early stress concentration, which has a very good warning effect.The experiments were operated under the fatigue load. The experiment materials are prefabricated with two typical defects which are incomplete penetration and slag inclusion. The signals were extracted through MMM technology and then analyzed.It’s been found out that the traditional damage criterion of the MMM technology has limits in hidden damage identifying. In order to solve the problem, the MMM signal,the magnetic filed vector space figure, Lissajous, and MMM characteristic extraction are used to analyze the incomplete penetration and slag inclusion defect. In this way,the results are obtained from qualitative analysis to quantitative analysis and the hidden damage criterion is given.It’s been found out that the MMM testing is susceptible to interference. The MMM signals fluctuate greatly and lack data samples. In order to predict the defect state using MMM technology, the gray theory is introduced. The prediction model of Kr is established based on GM(1,1) model and unbiased GM(1,1) model. The gray unbiased Markov chain model is obtained by connecting gray theory and Markov chain for improving the prediction precision and accuracy.The typical defects of welding specimens are studied by using the MMM testing first and then the gray unbiased Markov chain model is established, which provides a new tool for MMM technology identification.