Approximation theory

Daghestan Electronic Mathematical Reports, Issue 2 (2014)


On the identification of the parameters of linear systems using Chebyshev polynomials of the first kind and Chebyshev polynomials orthogonal on a uniform grid

UDK: 517.538

Pages: 1 - 32


Linear system in which the input signal $y = y(t)$ and the output $x = x(t)$ are related by the equation $x^{(r)}(t)=\sum_{\nu=0}^{r-1}a_\nu(t)x^{(\nu)}(t)+\sum_{\mu=0}^s b_\mu(t)y^{(\mu)}(t)$ is considered. The goal is to find the unknown variable coefficients $a_\nu(t)$ $(\nu=0,\ldots,r-1)$ and $b_\mu(t)$ $(\mu=0,\ldots,s)$ in case when the signal values are given in the nodes of a uniform grid $\Omega_N=\{t_j=-1+jh\}_{j=0}^{N-1}$, where $h=\frac2{N-1}$. It is assumed that the values of $x(t)$ and $y(t)$ are obtained as a result of experimental observations and are noised. For pretreatment of discrete information we apply <<anti-aliasing>> based on the use of Chebyshev polynomials orthogonal on a uniform grid $\Omega_N$. On the next step we switched from the original equation to the dual equation by representing of all figuring there functions (including derivatives) in the form of series by Chebyshev polynomials of the first kind $C_n(t)=\cos{(\arccos{t})}$. The result is a system of linear equations for the Fourier – Chebyshev coefficients of $a_\nu(t)$ and $b_\nu(t)$. Solving this system numerically, we obtain the variable coefficients of the original system of equations, thus completing the solution of the identification problem.


Keywords: Chebyshev polynomials of the first kind; Chebyshev polynomials orhtogonal on uniform grid; linear systems; signal processing; identification problem.




To issue content

Download full text