Study On Gearbox Fault Diagnosis Based On Empirical Wavelet Transform

Operating safety of machinery is directly affected by gearbox,which is the rotating machinery’s core component.Due to the harsh installation environment and the complicated internal structure,the parts,especially bearings and gears,are prone destroyed.It is practical significance for studying condition monitoring and fault diagnosis of gearbox,which can avoid major accidents.Empirical wavelet transform(EWT),particle swarm optimization(PSO)and maximum correlated kurtosis deconvolution(MCKD)were integrated by this paper,which successfully picked up rolling bearings’ faults from the measured signals.Constrained independent component analysis(CICA),which successfully separated the bearing and gear faults characteristics and completed the fault identification from the single-channel gearbox compound faults vibration signal.The main research was as follows:Firstly,the internal structure of the gearbox and bearings and gears’ vibration mechanism and common failure modes were explanationed.The fault signals of bearings and gears were imitated.In addition,the empirical mode decomposition,ensemble empirical mode decomposition,and variational modal decomposition were compared with EWT in simulation signals and bearing signals’ decomposition methods,components and time.This paper verified advantages in the decomposition ability and anti-modal aliasing of EWT.Secondly,the strong noise reduction ability of the MCKD was proved by the simulation signal.Aiming at the drawbacks of MCKD that the parameters need to be seted manually,this paper proposed an optimization algorithm based on PSO-MCKD,and successfully realized the adaptive parameter selection of MCKD.In addition,the method of selecting characteristic parameters based on kurtosis and cross-correlation coefficient was discussed.The bearing vibration signal was decomposed by PSO-MCKD-EWT,which successfully identifies the bearing fault information and realizes the effective diagnosis of early weak bearing fault.Finally,the basic principles of Independent Component Analysis(ICA)and CICA were introduced.The strong unmixing ability of the CICA was proved through comparative simulation analysis.Aiming at the compound faults of the gears and bearings in the gearbox,this paper proposed the PSO-MCKD-EWT-CICA algorithm,which can effectively separated the rolling bearing faults and gear faults.This algorithm provide new ideas for related fault diagnosis research.