Fault Degree Identification Of Rolling Bearing Based On QPSO-HMM

In modern manufacturing industry,more and more industrial production depends on large-scale mechanical equipment.Rolling bearing plays an important role in these mechanical equipment.It can be said that the state of rolling bearing directly determines the working efficiency of mechanical equipment.In order to make the large-scale rotating machinery keep reliable operation and high efficiency under the condition of long time and heavy load,accurate monitoring and fault diagnosis of bearing is of great significance.In this paper,the fault degree identification of rolling bearing is studied.In the whole research process,the feature extraction,dimension reduction and fault classification of rolling bearing fault vibration signal are deeply studied,and according to the hidden Markov model used for fault classification Model(referred to as HMM)in parameter estimation of the shortcomings of the quantum particle swarm algorithm(QPSO)Optimization(QPSO)and hidden Markov model are combined,and then fault diagnosis is carried out.Finally,the feasibility of this method is verified by the measured vibration signal of rolling bearing,and the comparison with Baum Welch parameter estimation method shows the advantages of QPSO-HMM proposed in this paper.The specific research contents are as follows:(1)The empirical mode decomposition(EMD)and variational mode decomposition(VMD)are compared by using the simulation signal and the measured signal.According to the comparison results,the better VMD is used as the signal decomposition method.(2)A feature extraction method of rolling bearing based on VMD-HT is proposed.The intrinsic mode function(IMF)is obtained by decomposing VMD,The instantaneous energy matrix is obtained by Hilbert transform(HT).The dimension of high-dimensional instantaneous energy matrix is reduced by singular value decomposition(SVD).The obtained singular value is taken as the vibration characteristics of rolling bearing,and the vibration characteristics are quantified for subsequent fault classification.(3)According to the problem that Baum Welch used in HMM parameter estimation is easy to fall into local optimum,QPSO is combined with HMM to estimate HMM parameters.This method makes use of the powerful global search ability of QPSO to effectively solve the problem of HMM falling into local optimum when calculating the initial matrix,state transition matrix and state emission matrix.(4)A method of rolling bearing fault diagnosis based on VMD-HT-QPSO-HMM is proposed.The vibration signal of rolling bearing is decomposed into several IMF by VMD,and the instantaneous energy matrix is obtained by HT transformation of IMF.The high-dimensional instantaneous energy matrix is reduced by using SVD,and several singular values are obtained.These singular values are used as the characteristic vectors of rolling bearing fault.Then different eigenvectors are quantized to get the sequence of eigenvectors.The vector sequence is input into QPSO-HMM for model training,and QPSO-HMM model for different fault states is established.After processing the test signal,the model is input into the trained model for rolling bearing fault diagnosis.Finally,the effectiveness of the proposed method is proved by the open vibration signals of rolling bearings in Case Western Reserve University,and the advantages of this method are proved by comparing with other methods.The results show that the method proposed in this paper can accurately diagnose the fault of rolling bearing.