Production Acceptance Testing Using an Improved Acoustic Signature Analysis

This paper presents the composite exceedance method as an alternative to the more costly broadband power spectral density (PSD) analysis which is the standard method of acoustic signature analysis for rotating machinery. The composite exceedance is superior to the PSD analysis in that the computational procedure is straightforward and effective without the use of Fourier Transforms and, therefore, lends itself to simple programming, and the information extracted from the vibration data is represented in one master plot, permitting direct evaluation of all g peak levels encountered in terms of the number of occurrences above which they exceed a given level of acceleration. The sensitivity of anomaly detection by the proposed method has been demonstrated in a number of actual cases, one of which is discussed. The data collection methodology and philosophy of the method as they relate to narrow-band Gaussian random processes are also considered.