A parametric approach for the estimation of the instantaneous speed of rotating machinery

A parametric method is proposed for the estimation of the instantaneous speed of rotating machines. The method belongs typically to the class of eigenvalue based parametric signal processing methods. The major advantage of parametric methods over frequency domain or time-frequency domain based methods, is their increased resolution and their reduced computational cost. Moreover, advantages of eigenvalue based methods over other parametric methods include their robustness to noise. Sensitivity analysis for the key parameters of the proposed method is performed, including the sampling frequency, the signal length and the robustness to noise. The effectiveness of the method is demonstrated in vibration measurements from a test rig during start-up and run-down, as well as during variations of the speed of a motorcycle engine. Compared to the Hilbert Transform and to the Discrete Energy Separation Algorithm (DESA), the proposed approach exhibits a better behavior, while it simultaneously presents computational simplicity, being able to be implemented analytically, even online. ► A Harmonic Decomposition parametric method is proposed for the IF estimation. ► HARD's major advantages include robustness to noise and low computational cost. ► The proposed method can track efficiently either large or slight speed variations. ► Compared to Hilbert transform and DESA, HARD exhibits a better behavior. ► Motor and motorcycle engine vibrations verify the efficiency of the HARD approach