Analysis of the multiple poles in the Vold-Kalman Filter

Vibration and noise signals coming of performing measurements in rotating systems - like reciprocating engines, transmissions, compressors, and pumps, for example - operating in non-stationary condition present some harmonic components that are rotational speed dependent, which are commonly know as orders. It is well known that if one intends to isolate one or some components of these signals, it is not recommended to use the traditional Fast Fourier Transform - if the analysis is in the frequency domain - or a fixed frequency band filter - to study it in the time domain. In these cases, it is important the use algorithms capable of treating those signals properly. This paper deals with the utilization of the Vold-Kalman filter to extract or eliminate a specific order component. This type of filter is able to track a time variant component using a reference signal - usually the angular displacement or velocity of the shaft. In this work, the influence of the number of poles of the filter on the quality of the filtering is verified through numerical simulations and the computational implementation of the filter aiming the manipulation of large block size is discussed.