Title :
Identification of rational approximate models in H/sup /spl infin// using generalized orthonormal basis
Author :
Szabo, Zoltan ; Bokor, Jozsef ; Schipp, Ferenc
Author_Institution :
Comput. & Autom. Inst., Hungarian Acad. of Sci., Budapest, Hungary
Abstract :
Robust identification with FIR models fails to be successful when the number of coefficients to be estimated becomes large, caused by lightly damped modes of the plant or poles very close to the unit circle. The paper presents a two-stage algorithm to obtain a low-order approximate model in frequency domain in a generalized orthonormal basis with guaranteed H ∞ error bound for deterministic linear time-invariant stable systems, the first stage being an L ∞ rational approximation and the second nonlinear step being an H ∞ rational approximation.
Keywords :
H/sup /spl infin control; approximation theory; frequency-domain analysis; identification; robust control; FIR models; H/sup /spl infin// rational approximation; L/sup /spl infin// rational approximation; LTI systems; deterministic linear time-invariant stable systems; generalized orthonormal basis; guaranteed H/sup /spl infin// error bound; lightly damped modes; low-order approximate model; poles; rational approximate model identification; robust identification; two-stage algorithm; Approximation algorithms; Finite impulse response filter; Frequency domain analysis; Noise level; Noise measurement; Robust control; Robustness; Sampling methods; Transfer functions; Uncertainty;
Journal_Title :
Automatic Control, IEEE Transactions on