Research On Non-destructive Measurement Of Hardness In Bearing Rings Based On Eddy Testing And Barkhausen Noise

As the main bearing part,the condition of the bearing will directly affect the performance,life and reliability of the machine.The main cause of bearing failure is surface wear,corrosion or high temperature oxidation caused by the surface performance of the outer ring material.In order to ensure that all bearing rings used in machinery and equipment meet the requirements for use,it is necessary to achieve a comprehensive,non-destructive,accurate classification and prediction of their hardness.Taking the bearing ring hardness testing as the background,this paper proposes a fusion of Eddy Current(EC)and Magnetic Barkhausen Noise(MBN)data based on the inaccuracy of the current hardness classification and prediction of heat-treated test pieces.The hardness testing method proposed in this paper combines the advantages of EC and MBN testing to improve the accuracy of non-destructive classification and testing of hardness.The main research contents of this article are as follows:(1)Starting from the process and characteristics of the heat treatment process,the material of the heat treated bearing ring and the main reasons for the change in hardness are introduced,which lays a theoretical research for the experiment and further research.(2)Design and carry out Leeb hardness measurement experiments,eddy current detection experiments and Barkhausen noise detection experiments.Obtain the Leeb hardness value of each test point as the hardness standard value.The eddy current detection signal and Barkhausen noise detection signal are obtained,and the characteristic values are obtained through data processing to provide data samples for hardness classification and quantitative prediction.(3)An intelligent nondestructive classification method for hardness of heat-treated bearing rings is proposed.First,the effect of eddy current and Barkhausen noise eigenvalues on the hardness classification of the test pieces is analyzed.It is found that the eddy current and Barkhausen noise detection eigenvalues cannot accurately distinguish the three types of test pieces,but the eddy current high-frequency reactance signal and the square value,mean value,peak value,peak time of Barkhausen can accurately identify the unqualified and unheated test pieces respectively.Data fusion of the eigenvalues of the two methods results in four fusion characteristics of RMS-X,mean value-X,peak value-X,peak time-X,which can effectively distinguish the three types of specimens,And use softmax classification algorithm for classification.The results show that the "RMS-X" softmax classification model performs best in applicability and classification accuracy,and can achieve accurate classification of qualified,unqualified and unheated specimens.(4)A hardness testing method for heat-treated bearing rings based on BP neural network is proposed.By comparing the BP regression effects of the eddy current and Barkhausen noise characteristics,it is found that the Barkhausen mean square value,mean value,peak value,and peak time perform well in the hardness test.The eddy current resistance value at 1kHz can also be used in the hardness test.After that,the joint hardness test is performed using the characteristic values obtained by the test.The results show that the BP neural network trained by the mean value,mean value,peak value and eddy current resistance at 1kHz characteristics of the two detection methods performs best,with a maximum absolute error of 5.6 and a maximum relative error of 1.13%.The detection accuracy meets Engineering requirements.The paper has 49 pictures,21 tables,and 93 references.