Fault detection and diagnosis of a multistage helical gearbox using magnitude and phase information from vibration signals

Gu, James Xi (2016) Fault detection and diagnosis of a multistage helical gearbox using magnitude and phase information from vibration signals. Doctoral thesis (PhD), Manchester Metropolitan University.

Preview

Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (6MB) | Preview

Abstract

Vibration generated by a gearbox carries a great deal of information regarding its health condition. This research aims primarily on the detection and diagnosis of tooth defects in a multistage gearbox based on advanced vibration analysis. Time synchronised averaging (TSA) analysis is effective at removing noise but it is inefficient in implementation and in diagnosing different types of faults such as bearing defects other than gears. Conventional bispectrum (CB) can eliminate Gaussian noise while it preserves the signal’s phase information, however its overpopulated contents can still provide inaccurate information regarding to different types of gear faults. Recently developed modulation signal bispectrum (MSB) has the high potential to lead to the high accuracy of diagnostics of gearboxes as it more effectively characterises modulation signals such as gearbox vibrations. Therefore, the research takes MSB as the fundamental tool for analysing gearbox vibration signals and developing accurate diagnostic techniques. Firstly, it has realised that conventional techniques often ignore the effect of phase information in gearbox diagnostics. This thesis then focuses on developing CB and MSB based techniques for detecting and diagnosing of gearbox faults. Secondly, it has found that vibration responses from a multiple stage gearbox have high interferences between amplitude modulation (AM) and phase modulation (PM) which can be formalised from both gear faults and inherent manufacturing errors. However, the faults can induce wider bandwidth vibrations. Correspondingly, optimal component based schemes are also developed based on the use of MSB coherence results. Then the proposed MSB method allows an effective gearbox diagnosis using the signals in a narrower frequency band that is below twice the rotational frequency plus the highest meshing frequency amongst different gear transmission stages, being more suitable for wireless network condition monitoring systems. It has also found that the signals at resonance frequencies has a higher signal-to-noise ratio and more effective for obtaining accurate diagnosis. Also software encoder based TSA was found to be not robust and accurate due to the influences of noise and referencing components on obtaining a reliable phase signal for implementing TSA. Finally, the diagnostics carried out upon different fault cases using both CB and MSB have verified the proposed approaches can provide accurate diagnostic results, and with the new MSB based detector and estimator being more effective in differentiating between diffident fault locations for two local and one non-uniformly distributed tooth damages in a two stage helical gearbox.