Noise subtraction and marginal enhanced square envelope spectrum (MESES) for the identification of bearing defects in centrifugal and axial pump

Abstract The natural intermittent impulses created by impeller in case of centrifugal pump, and reciprocating motions of piston in case of axial pump exhibit strong cyclo-stationary phenomenon. This makes the identification of defect impulses difficult in their vibration signals even with the powerful signal processing techniques such as Fast-spectral correlation. To deal with the issue, this work proposes noise subtraction and marginal enhanced square envelope spectrum (MESES) for detecting bearing defects in the centrifugal and axial pump. For subtraction of noise, Fast Fourier Transform (FFT) of signal of unknown defect condition is computed and then FFT of normal working condition is subtracted from the signal of unknown defect condition. Then, inverse FFT is applied to resulting signal to construct denoised signal in time domain. The noise subtraction is done in frequency domain so as to avoid time lag problem which generally occurs in two signals obtained at different time. Further, the signal is processed by Fast- spectral correlation. To select the spectral frequency having fault related information, a criterion named as marginal band selection indicator (MBSI) is proposed. The spectral frequency band having the highest MBSI is selected for computing MESES. After selection of spectral frequency, MESES is computed and Feature frequency is find out. Finally, comparison of feature frequency is done with the bearing fundamental fault frequencies to conform the presence of defect.