Vibration based centrifugal pump fault diagnosis based on modulation signal bispectrum analysis

This paper characterises vibration signals using modulation signal bispectrum method in order to develop an effective and reliable feature sets for detecting and diagnosing faults from both the bearings and impellers in a centrifugal pump. As vibration signals contain high level background noises due to inevitable flow cavitation and turbulences, effective noise reduction and reliable feature extraction are critical procedures in vibration signal analysis. Considering the modulation effect between rotating shaft and vane passing components, a modulation signal bispectrum (MSB) method is employed to extract these deterministic characteristics of modulating components in a low frequency band for diagnosing both the bearing defects and impeller blockages. Experimental results show that the diagnostic features developed by MSB allow impellers with inlet vane damages and bearing outer-race faults to be identified under different operating conditions. Not only does this new method produces reliable diagnostic results but also it needs a bandwidth about 1000Hz, rather than the high frequency bands around 10kHz used by conventional envelope analysis.