Application Research Of Wavelet-Based Demodulation In Pump Fault Diagnosis

The running status of dry vacuum pump influences the stability of the whole vacuum system and the hydraulic pump is the"heart"of the hydraulic system. So the condition monitoring and fault diagnosis of the rotary machine such as dry vavuum pump and hydraulic pump is crucial.This paper advances the extractive technique as subject investigated,A Morlet-like wavelet cluster-based method for band-pass filtering and envelope demodulation is described and it is applied for fault diagnosis relating to the rotary machine such as dry vavuum pump and hydraulic pump.It is not only important for theoretical study,but also has popularization to other rotary machine.This paper summarizes vibration monitoring technique relating to rotary machine,then illustrates how to select the testing position and the vibrational sensor.Especially makes profound institute to amplitude domain and time domain and frequency domain analysis method.Systematically researches the basic ideas of Wavelet Transfomation theories when it is used in the vibrational signals.It is verified that wavelet analysis method is unique compared to FFT. The method is suitable for most rotary machine.When using classical wavelet envelope demodulation method, generally first makes wavelet decomposition for filtering to the vibration signals ,then Hilbert Transformation,Last we can get the envelope signal .In this way the process of calculation is complex. Here a Morlet-like wavelet cluster-based method for band-pass filtering and envelope demodulation is described. Via appropriate choice of wavelet parameters, a wavelet cluster combined with multi-Morlet-like wavelet can be used as a band-pass filter with zero phase shift, flat topped pass-band and rapid attenuation in the transition band. It can be used to extract high frequency natural vibration components. This new method is applied for fault diagnosis relating to the rotary machine such as dry vavuum pump and hydraulic pump. It is shown that the fault characteristic frequencies can be extracted effectively. The efficacy of the method is demonstrated in experimental studies.