Research On Material Damage Monitoring Technology Based On Acoustic Emission

In order to monitor the damage of high-strength materials,the static test of titanium alloy machine and the tensile test of glass fiber reinforced resin matrix composites were carried out.Acoustic emission technology was used to obtain the acoustic emission signals of different materials during the damage process,and combined with metal and Acoustic emission parameters were analyzed for the damage mechanism of the composite.In terms of composite materials,the acoustic emission activities of glass fiber reinforced resin matrix composites at different loading rates are significantly different.Combined with the load curve of the test piece,the damage of the material can be reflected by parameters such as acoustic emission energy,RMS value,and ringing count.To the extent,the damage type of the composite can be reflected by the change of the peak frequency of the acoustic emission,and the tensile damage data of the glass fiber resin material is segmented according to the degree of damage and the type of damage.In terms of metal materials,the activity of static acoustic emission signals is described in combination with the plastic deformation mechanism and microscopic mechanism of titanium alloy.The titanium alloy exhibits two peaks of acoustic emission activity under static load,according to the parameters such as absolute energy of acoustic emission.The damage data is segmented by the distribution of the sum signal during loading.Manually judging material damage requires a combination of rich material knowledge and the need to obtain changes in the material over time.In order to perform real-time intelligent monitoring of equipment materials in service and introduce deep learning methods,this paper introduces the material damage phase based on deep belief network.The identification method mainly includes 4 steps.Firstly,the acoustic emission signal of the material loading damage is obtained by the acoustic emission sensor;then the damage stage is divided according to the material properties and mechanical properties,such as according to the yield point or damage mode of the material;then the label is marked by the deep belief network.The acoustic emission data is trained to obtain the damage phase identification network.Finally,the trained deep belief network model is used to identify the damage of the test set or the data to be diagnosed.Finally,the accuracy of the identification data of the tensile test of the composite material reached 98%,and the accuracy of the titanium alloy static load damage test data reached 80%.