Gear Fault Diagnosis And Features Extraction Based On Envelope Synchronous Average

Gear is one of the most important parts in rotating machinery for changing speed or transmitting power. However, it is also one of the parts where faults often occurred in rotating machinery. Due to the complex transmission path and the fact that the vibration source is far from the sensor mounted positions, the vibrations of gears are difficult to analyze and the accurate determination of the fault is not trivial. These facts may affect the production when faults are serious and even trigger unsafe operations for workers. In addition, most of available studies of gear faults are mainly focused on the vibration signals under the constant running speed, and thus lack non-stationary fault feature extraction techniques, e.g. the run-up or coast down stages of the rotating machinery. Consequently, it is necessary to further develop new technology for non-stationary gear fault diagnosis.Vibration analysis is usually used in faults diagnosis of machinery. As we know, the fault could be determined by the effective vibration analysis method. Recently, some signal processing technologies have been proposed and developed in this field, so that the gear fault features can be extracted from the picked up vibration with complex background noise. This has become the research forefront of gear fault diagnosis, and thus will be presented and studied in this thesis.With a deep study and research on typical gear faults and an extensive review of the achievements of previous researchers, the gear faults diagnosis based on envelope synchronous average has been proposed in this thesis. In the proposed approach, the fast kurtogram algorithm is utilized to obtain the optimal parameters of resonance demodulation, to extract the complex envelope signal of gear adaptively and to highlight the shock components at first. Then the complex envelope signal is converted into angular domain, where the real and imaginary of the complex envelope angular signal are processed by the synchronous average in the angular domain for different reference shafts. Finally, the order spectrum analysis is employed to extract gear fault features.Meanwhile, this thesis also focused on the fault features extraction of gears under variable speed operations. During the run-up or coast down, the computed order tracking is employed to covert the time domain signal into angular domain for the gear fault features extraction with variable speed. Furthermore, we combine the envelope analysis and the synchronous average in angular domain as proposed by other researchers. By this way, we can obtain the gear fault features for different shaft through choosing different shafts as the reference shaft in the synchronous average technology, and the multi-gear failures that are usually difficult to be separated can also be resolved effectively. Consequently, it is shown that broadband noises can be eliminated effectively, the gear fault order components can be extracted from original vibration signals, and the frequency blur caused by the speed fluctuations can also be canceled at the same time.Following above theoretical studies, simulation and actual test experiments were all conducted to verify the validity of the proposed method.