Estimation of instantaneous speed for rotating systems: New processing techniques

Abstract

One of the most important parameters in the analysis of rotating systems is the instantaneous speed or RPM. The estimates from a tachometer data generally contain rapid unacceptable fluctuations. A new technique called the Adaptive nth Pulse Algorithm is presented in this paper which addresses the fluctuations in values from the conventional RPM estimation procedure through a root cause analysis. In addition to significantly improving the signal to noise ratio of the curve, this method has the advantage of transforming the rotation speed from being a function of angle of rotation to a function of time. An alternative approach has been developed by B&K called the Autotracker Algorithm where instantaneous speed is estimated using Bayesian statistics from measured accelerometer or microphone data thereby avoiding the need for tachometers. In the current paper, a smart algorithm built upon this Autotracker Algorithm has been developed which is more robust and as much as 100 times faster. Analysis of data with crossovers between orders and dominant natural frequencies is possible using a reprocessing step in this algorithm. Data with very high slew rate and speeds reaching 25, 000 RPM has been shown to be processed in under 5 sec. Comparison is made between the Bayesian and nth pulse estimates for several data sets to show that the results are consistent with each other. Examples include data from a rotor rig, a 6 – cylinder engine powertrain and a vacuum cleaner.