Quadratic Phase Coupling Detection And Its Applications In Locomotive Fault Diagnosis

This thesis investigates the quadratic phase coupling (QPC) of locomotive signal and the mechanisms giving rise to quadratic phase coupling. Then puts the focus on the relationship between locomotive faults and QPC. Moreover, the modified bispectrum and empirical mode decomposition are introduced to fault feature extraction. The main contributions and conclusions are made as follows:1. The conventional methods of quadratic phase coupling detection are introduced. In the lights of the nonlinear properties of locomotive signals, the mechanisms of gear, bearing and motor giving rise to nonlinear coupling are discussed. Furthermore, a framework for analyzing machinery faults is established by means of quadratic phase coupling detection.2. When faults emerge in rolling element bearing, the vibration signal will display more nonlinear properties than normal state. One of the nonlinearity comes from that some harmonic components may interact one another and contribute to the power at the sums and/or differences of their frequencies. The modified bispectrum was applied to identify these interactions and detect the bearing fault while it is still in an incipient stage.3. EMD is a fully data-driven signal processing method and acts essentially as a filter bank. The bearing vibration signal is adaptively decomposed into intrinsic oscillatory components called intrinsic mode functions (IMFs), with each IMF corresponding to a specific range of frequency components contained within the signal. Thus, the noise can be suppressed effectively.4. Based on the investigations mentioned above, a fault diagnostic method by jointly using EMD and modified bispectrum is developed. The proposed approach was effectively applied to the operation tests of SS7E type locomotive, and the bearing fault of traction motor was diagnosed successfully.