Approximate identification in Laguerre and Kautz bases (Q1295141)

The paper is devoted to a specific consideration on the use of specific basis functions and the construction of a class of linear approximation algorithms. These algorithms are generated on the base of models, which are weighted partial sum operators in the Laguerre and Kautz bases. A class of linear approximation algorithms generated by \(\varphi\)-weighted partial sums of a biorthogonal expansion in the Laguerre and Kautz basis is under consideration. The problem studied is the first step of the two-step identification under an \(H^\infty\) criterion. Bounds on the partial sum of operators and on the \(L^\infty\) norm of the approximation error are derived. A frequency domain identification procedure is developed where discrete Fourier transformation of appropriately transformed data computes the model parameters.