Identification of a time-varying parameter of a noiseless sinusoidal signal (Q2171753)

This paper considers the problem of identifying an unknown time-varying parameter \(\omega(t)\) of a noiseless sinusoidal signal of the form \[ y(t) = \alpha(t) \sin(\omega(t) + \varphi), \] where \(y(t)\) is a directly measurable signal, \(\alpha(t)\) is an unknown amplitude, \(\omega(t)\) is an unknown time-varying parameter, and \(\varphi\) is an unknown constant phase shift. It is assumed that \(\alpha(t)\) and \(\omega(t)\) are solutions of linear time-invariant differential equations with known coefficients but unknown initial conditions. A new algorithm for estimating \(\omega(t)\) is proposed by using gradient tuning algorithms based on a linear regression equation obtained by parametrizing the original parameter-nonlinear sinusoidal signal. An example is given to explain the procedure for the synthesis of the estimation algorithm, and to show the efficiency of the proposed algorithm.