Title :
An adaptive algorithm for identification of IIR systems
Author_Institution :
Dept. of Mech. Eng. & Appl. Mech., Michigan Univ., Ann Arbor, MI, USA
Abstract :
An adaptive identification algorithm formulated with a weight matrix is proposed. The series-parallel model is used with tapped delay lines to represent a dynamic mapping problem. Hyland´s idea (1991) using a constraint matrix with Hadamard (element-by-element)-product to preserve the causality of dynamical systems is extended to IIR systems. By introducing an appropriate constraint matrix corresponding to the parameterization we can directly obtain the parameter without subsequent realization. The adaptive step size of the proposed algorithm is derived by using a projection technique. Stability and convergence of the proposed algorithm are investigated with appropriate assumptions. Simulation results are included to demonstrate the convergence of the proposed algorithm
Keywords :
convergence; identification; matrix algebra; stability; Hadamard-product; IIR systems identification; adaptive algorithm; adaptive step size; constraint matrix; convergence; dynamical systems causality; element-by-element product; series-parallel model; stability; tapped delay lines; weight matrix; Adaptive algorithm; Convergence; Delay lines; Equations; Erbium; Gaussian processes; Mechanical engineering; Neural networks; Recurrent neural networks; Stability;
Conference_Titel :
Decision and Control, 1993., Proceedings of the 32nd IEEE Conference on
Conference_Location :
San Antonio, TX
Print_ISBN :
0-7803-1298-8
DOI :
10.1109/CDC.1993.325666
