LAUSR

Abstract Fault diagnosis of planetary gearboxes under variable-speed conditions is a challenging task since the vibration signals are non-stationary and have more complicated characteristic components due to the complex gearbox configuration. Effectively identifying and extracting these non-stationary characteristic components are important for fault diagnosis. This paper achieves the goal by exploiting an improved adaptive chirp mode decomposition (I-ACMD) method. The I-ACMD mainly includes two ingredients. Firstly, the algorithm framework of the original ACMD is modified for joint component estimation, which can effectively deal with very close signal components. Secondly, to address the issue of the instantaneous frequency (IF) initialization for I-ACMD, we combine a parameterized demodulation (PD) method with a signal resampling technique to extract multiple IF curves simultaneously. Compared with traditional time-frequency ridge detection methods, the proposed PD-based IF initialization method shows much better interference robustness and thus is more effective to analyze complicated vibration signals of planetary gearboxes. Moreover, with the output results of the I-ACMD, we construct a high-resolution time-frequency representation which can clearly reveal the time-varying gear characteristic frequencies. Both our simulated and experimental studies have shown that the proposed method can effectively indentify very close and weak vibration characteristic components, and thus successfully detect different kinds of gear faults.