INSTANTANEOUS FREQUENCY ESTIMATION IN ROTATING MACHINERY USING A HARMONIC SIGNAL DECOMPOSITION (HARD) PARAMETRIC METHOD

摘要

A Harmonic Signal Decomposition (HARD) parametric method is proposed for the estimation of the instantaneous speed of rotating machinery, which makes use of vibration signals measured on a machine submitted to relative large speed variations, such as those occurring during start up or run down. The HARD method belongs typically to the class of eigenvalue based harmonic signal detection methods. The major advantage of the method over other proposed approaches is the accuracy offered by the eigenvalue decomposition methods and the computational efficiency, which may lead even to an on-line implementation of the method. The way to select the critical parameters of the method, I.e the sampling frequency and the signal length, are discussed. The performance of the method is evaluated in sinusoidal and chirp simulated signals under various noise environments, as well as to vibration measurements during the start-up and run-down of an industrial pump. As shown, the method is relatively simple to implement, has a simple computational implementation and can track efficiently the speed variations of rotating machinery.