Singularity recognization of vibration signal based on Hermitian wavelet for diagnosis

The vibration signal in the nature of singularity is always caused by mechanical fault of equipment. It is important to recognize the signal singularity correctly for mechanical fault diagnosis. The complex Hermitian wavelet is constructed by means of the first and the second derivatives of the Gaussian function to detect signal singularities. The Hermitian wavelet has a real Fourier transform that will not mix the signal phase with its filter phase, and its real part and imaginary part have less oscillation than Morlet wavelet, so that the convolution operation can be process with fewer number of data points and the signal singularity will not be smeared. The time-scale amplitude plot and phase plot based on Hermitian wavelet are presented to detect signal singularities. A successful application has been obtained in diagnosis of large air compressor gearbox. According to the singularities extracted from amplitude plot and phase plot of gearbox vibration signals, it is obvious that there existed impact rub fault that caused intense vibration in the gearbox. After justified maintenance the compressor gearbox runs smoothly in a good condition.