Magnetic-Acoustic Feature Extraction And Damage Fusion Evaluation Of 45 Steel Components In Fatigue Process

Ferromagnetic components have high strength and toughness,which are widely used in construction machinery,power equipment,pressure vessels and other industries.Fatigue damage is a typical damage form of ferromagnetic components.Effective detection of fatigue damage is an important measure to ensure the safe service of ferromagnetic components.In recent years,due to the sensitivity of magnetic memory and acoustic emission characteristics to the early damage of components and the real-time monitoring,magnetic and acoustic characteristics are increasingly used in the fatigue damage assessment of ferromagnetic components.In this paper,the magnetic acoustic feature extraction and damage fusion evaluation of45 steel samples in fatigue process are carried out,which can realize the mutual verification of multi feature quantity in the damage detection process,make up for the deficiency of single feature quantity detection,and provide the basis for the quantitative evaluation of fatigue damage of ferromagnetic components.In order to clarify the change law of magneto acoustic characteristics in fatigue damage process of ferromagnetic components,an experimental platform for the detection of magnetic memory and emission of fatigue process was built.Through the low cycle fatigue experiment,the magnetic memory and acoustic emission signals under different cyclic loading cycles are collected,and the variation rules of magnetic and acoustic signals are analyzed.The multi-scale entropy characteristics of magnetic memory signal and the wavelet packet energy spectrum and singularity index characteristics of acoustic emission signal are further extracted.On this basis,two methods,support vector machine and plain Bayesian,were selected to construct a fusion model for the evaluation of the magnetic-acoustic features of the fatigue damage process.The results show that the mean value of the multiscale entropy of the normal magnetic field intensity H_p(y)gradually increases and the mean value of the multiscale entropy of the magnetic field intensity gradient K of the magnetic memory signal gradually decreases as the degree of fatigue damage(number of cycles)increases.The cumulative ringing count value of the acoustic emission signal gradually decreases with increasing number of fatigue cycles,and the singularity index of the energy(1 band share)and acoustic emission amplitude gradually increases with increasing number of fatigue cycles.A comparative analysis of the models shows that the prediction accuracy of the magnetic-acoustic feature fusion model using the plain Bayesian approach is higher.The results of this paper can provide method and model support for fatigue damage assessment of ferromagnetic components,and have a certain industry application prospect.Figure[32]table[9]reference[88]...