Bearing Fault Diagnosis Based On The Phase Space Reconstruction Of The Maximum Joint Entroy And Recurrence Quantification Anaalysis

Bearing is used as the broad generic component in rotating machinery, whose defectsand damage will directly impact the stable operation of the equipment, even cause thedamage of the entire equipment. Thus, the fault diagnosis and identification of bearing isparticularly important. However, due to the nonlinear and non-stationary characteristics ofthe bearing fault signals, it limits the traditional fault feature extraction method establishedon the basis of signal stability. Aim at the complex features of mechanical failure vibrationsignals, this paper presents a bearing fault diagnosis method based on the phase spacereconstruction of the maximum joint entropy and recurrence quantification analysis.Meanwhile, it uses GG fuzzy clustering algorithm to identify the status of fault signal.And the above theoretical research is applied in the rolling bearings and rotor faultdetection.Firstly, the common calculation methods of vibration fault and fault frequency isdescribed in this paper and meanwhile the conventional fault diagnosis methods: timedomain analysis and frequency domain analysis is included.Secondly, the obvious advantage of the maximum joint entropy criterion of optimaldelay time and the symbolic analysis method for calculating maximum joint entropy isanalyzed. Then a new delay time obtaining method by the maximum joint entropy basedon the symbolic analysis is proposed. The best embedding dimension is acquired by Caomethod. Numerical experiments show that the method can reconstruct the original phasespace accurately, rapidly and efficiently.Thirdly, on the basis of accurate calculation of the phase space reconstructionparameters, the recurrence plot method is introduced, which can qualitative representdynamical properties of the system graphically. The recurrence quantification analysis(RQA) method is able to quantify the recursive phenomenon shown in recurrence plot.The RQA method is applied to diagnose mechanical fault. The main nonlinearcharacteristics are extracted to compose the vector of bearing fault identification.Combined with the GG fuzzy clustering, it can achieve the fault pattern recognition of bearings.Finally, the datasets of the rolling bearing fault from the Case Western ReserveUniversity and the rolling mill data measured from Baosteel1580SP are taken as theexperiment research object. It selects the maximum joint entropy method to get theparameters of phase space reconstruction, and uses the RQA and GG fuzzy clusteringalgorithm to diagnose and identify fault signal. The results demonstrate that these methodsare able to achieve the diagnosis and identification of the rolling bearing fault and thesignal of transmission system bearing.