Precise order tracking analysis of time-varying vibro-acoustic signature from rotating machines

Abstract

Transient operation of rotating machines usually exhibit a large variation in rotating speed and frequencies of interest often fluctuate due to RPM variations. Order tracking is useful for the transient analysis of vibro-acoustic signal from a rotating machine. A fine resolution is required both in time and frequency domains to trace the spectral magnitude of the interested frequency in precision. In this study, characteristics of popular order tracking methods using short time Fourier transform (STFT) and Vold-Kalman filter are studied in the viewpoint of calculation time and resolution. They are compared with a new tracking method using variable frequency resolution STFT (VFR-STFT), being suggested for the sound quality analysis. For a test signal, vehicle interior noise measured in an idle run-up condition is used. Under the same signal processing conditions, time and frequency domain resolutions produced by three methods are compared. It is observed that VFR-STFT detects bandwidth of high order components of engine firing frequency precisely and it produced the finest resolution in time and frequency domain among three methods. Although VFR-STFT needs a bit more calculation time compared to the other methods, it is useful for the cases that do not require a real time analysis.