Adaptive Disturbance Rejection Using ARMARKOV/Toeplitz Models

Dennis Bernstein, Aerospace Engineering Department, University of Michigan

In this talk I will describe recent results on identification and adaptive control based upon ARMARKOV/Toeplitz models, which are time series models that explicitly incorporate impulse-response parameters. The adaptive disturbance rejection algorithm is based upon a sliding-window model and gradient update with guaranteed performance improvement. A useful feature of the algorithm is the fact that it requires an ARMARKOV model of only the secondary path (control-to-performance) dynamics. Experimental results in multivariable activ e noise control will be used to demonstrate the performance of the algorithm in the presence of tonal, multi-tone, broadband, and time-varying disturbance spectra without direct measurement of the disturbance. Robustness of the algorithm to uncertainties and nonlinearities in the secondary path dynamics will also be discussed.