Nonlinearity detection in hydraulic machines utilizing bispectral analysis

• Main Author: Mohamed Saeed, Somia Alfatih
• Format: Thesis
• Published: 2013
• Subjects: TJ Mechanical engineering and machinery
• Online Access: http://eprints.utm.my/id/eprint/42178/; http://dms.library.utm.my:8080/vital/access/manager/Repository/vital:78200

Higher order spectral analysis is one of the relatively more recent tools in signal processing used for detection and identification of higher harmonics in a signal. It is also acknowledged to be an effective tool for detecting nonlinear behavior in mechanical systems. In this study, vibration sources in piping system and in hydraulic machine which may have features of nonlinear behavior were investigated. An experimental study was undertaken to formulate a more sensitive and effective method using bispectral analysis to diagnose resonance in piping and cavitation in a centrifugal pump test facility. Data was obtained for normal and resonant conditions at different locations on the pipes, and analyzed using FFT and bispectrum methods. Bispectral analysis was shown to be a more effective tool in detecting coupled frequencies due to resonance. Cavitation was induced on the suction side of the pump. The cavitation signal was analyzed with and without induced cavitation conditions. It was observed that bispectral analysis could be used as an early indicator of cavitation with changes for severity of cavitation. Higher order spectral analysis was also undertaken on field data from compressor piping with suspected resonance. High frequencies random vibration was shown to be a good indicator of the excitation of one of the system component’s natural frequency. Field measurement results which required bump test confirmed resonance related fault. The results of bispectrum analysis showed the presence of nonlinearity in the spectrum due to the resonant condition without the need for a bump test. The findings suggested that bispectral analysis could be used as a diagnostic tool for machinery vibration related faults detection especially when plant shutdowns are not possible.