Detection of machine soft foot with vibration analysis

Soft foot is considered to be one of the main causes of vibration problems in rotating machinery. However, there have been relatively limited research efforts to develop robust diagnosis tools for the early detection of such problems, particularly in scenarios where the operational vibration background noise of the machine is high. All previous studies have utilised the Fourier spectrum to diagnose soft foot symptoms. The present study is aimed at developing a series of signal processing techniques to reduce the effect of vibration background noise whilst extracting the fault feature. Three signal processing techniques were applied: adaptive filter, spectral kurtosis and envelope analysis. All three techniques were applied to measured vibration signals acquired from a motor-compressor test-rig. In conclusion, it is shown that the techniques detected motor soft foot more effectively than the conventional spectral method.