Fault Diagnosis of Planetary Gearboxes Based on Maximum Correlated Kurtosis Deconvolution

Due to the structure of a planetary gear train is complicated and the fault signal feature is extremely difficult to extract, so the method of using torsion vibration signal to diagnose the fault of planetary gearboxes is proposed. Based on comparison of spectrums of reciprocating vibration signals and torsion vibration signal, finally the result that torsion vibration signals were more suitable for the fault diagnosis of a planetary gear train than reciprocating vibration signals was obtained. A fault diagnosis method for torsion vibration signal based on maximum correlated kurtosis deconvolution was proposed, because the strength of torsion vibration signals is weak and the impact characteristics are not obvious. Firstly the method of zero mean was applied to process torsion vibration signals of planetary gearbox, then MCKD method was utilized to enhance the impact characteristics of torsion vibration signals. The length of FIR filter is based on the kurtosis of fault impact characteristics. Eventually the impact characteristics of the torsion vibration signal of planetary gearbox received a significant promotion by MCKD. The method to reduce the noise component of the torsion vibration signal and improve the cycle impact fault characteristics is effective. It is appropriate for the fault diagnosis of torsion vibration signals of planetary gearboxes.