Order Spectrum Based on Generalized Demodulation for Gearbox Fault Diagnosis under Nonstationary Conditions  
Author Xiaowang Chen


Co-Author(s) Zhipeng Feng; Ming Liang


Abstract Extraction of time-variant characteristic frequency components is a key issue of gearbox fault diagnosis under nonstationary conditions. In the last decade, order tracking technique has been widely employed in nonstationary signal analysis and different methods for order spectrum analysis were proposed, but these methods mostly demand for either an encoder or resampling operation to acquire stationary signal in angular domain. To be free from extra hardware and evitable resampling errors, this paper proposes a novel method to extract order spectrum from nonstationary signals by generalized demodulation. The whole frequency band is divided into a series of shaft rotating frequency orders and generalized demodulation is performed iteratively to convert all timevariant order components into constant frequency ones. Then the order spectrum of original nonstationary signal is obtained as the superposition of Fourier spectra of all mapped order components. The effectiveness of proposed method is validated by both numerical simulated and lab experimental gearbox vibration signals under nonstaionary conditions. The detected modulation sidebands imply the effectiveness of proposed method in revealing time-variant features and diagnosing gear fault under nonstationary conditions.


Keywords order spectrum, fault diagnosis, nonstationary, generalized demodulation
    Article #:  22254
Proceedings of the 22nd ISSAT International Conference on Reliability and Quality in Design
August 4-6, 2016 - Los Angeles, California, U.S.A.