Current-Aided Dynamic Time Warping for Planetary Gearbox Fault Detection at Time-Varying Speeds

One of the main techniques for tacholess order tracking is to extract instantaneous frequency (IF) ridges from the time-frequency representation (TFR) of vibration signals of rotating machinery. However, the extraction process is hindered by limitations in both time-frequency resolution and the accuracy of ridge extraction algorithms. It is time-consuming and laborious. This article proposes an adaptive tacholess order tracking method with dynamic time warping (DTW) aided by the current signal for fault detection of planetary gearboxes. The approach overcomes the aforementioned challenges by first extracting the instantaneous fundamental frequency from the TFR of the current signal to obtain the shaft rotation frequency. Second, the shaft vibration signal at time-varying speed and the reference signal at constant speed are established. Then, the DTW algorithm is used to obtain the optimal warping path of the shaft vibration signal and the reference signal, which enables the resampling of the planetary gearbox vibration signal. Finally, the envelope order spectrum is employed to detect the planetary gearbox fault components. The effectiveness of the proposed method is verified by simulation and industrial-grade experiment bench.