An Order Demodulation Analysis Method for Planetary Gearboxes

In addressing challenges such as the speed change of planetary gearbox (PG) during operation and the complexity in extracting fault features, an order demodulation analysis method based on order tracking (OT) and variational mode decomposition (VMD) is proposed. Non-stationary time domain signals are transformed into relatively stable angular domain signals through angular resampling. The intrinsic mode function (IMF) is obtained by demodulation of the angular multi-component signal using the VMD method. The instantaneous order of a proper IMF is calculated via Hilbert Transform. The sensitive single component is selected by matching the meshing order and the central order of the instantaneous order. Finally, the amplitude of the characteristic order components extracted from the instantaneous order spectrum was used as features and Support Vector Machine (SVM) was employed for the classification of four different conditions in the sun gear. Comparing the results of the proposed method with Deep Neural Networks (DNN) and Convolutional Neural Networks (CNN), the experimental results demonstrate the effectiveness of the proposed method in diagnosing PGs under non-stationary conditions.