Frequency determination in control applications: excitation-based approach

Artikel i vetenskaplig tidskrift, 2012

New algorithms for estimation of the frequencies of oscillating waveform signals are described. A model of the signals is presented in the form of a linear difference equation with unknown coefficients, which define the frequencies and amplitudes. Coefficients are estimated utilizing the property of the persistence of excitation of oscillating signals. Exponentially damped and oscillating signals are described in a unified framework. A property of excitation is proved for an exponentially damped signal that contains a single frequency via diagonal dominance of an information matrix. Two applications of this frequency estimation technique are considered. The first one is filtering of the wind speed signal in wind turbine control applications, and the second one is the frequency estimation of exponentially damped signals motivated by the engine knock detection applications.