Analysis, approach and assessment of vibration criteria in shipboard machinery condition monitoring and diagnostics

Abstract

The setting of alarm levels plays a vital role in a machinery condition monitoring and diagnostic system. In this research, two approaches to setting vibration alarm levels using vibration signals produced by fire pumps are presented in the time and frequency domains. In the time domain, the cross peak analysis (CPA) is proposed to extract the dominate peak points. The distribution of these cross peak points is found to have a lognormal distribution and can be normalized to a Normal distribution in the VdB domain. The computed microns + 2 Sigma value in the VdB domain is suggested for use as the alarm level. In the frequency domain, the 1/1 octave band analysis (OBA) is introduced. Three artificial fault simulations are conducted to compare the 1/1 octave band method with the broadband method. The results show that the 1/1 octave band method is more sensitive to the changes in VdB level than the broadband method. The computer programs to perform these two analyses are written using MATLAB. Examples of the use of these programs are included in this report