Skip to main content
eScholarship
Open Access Publications from the University of California

Control relevant identification of plant and disturbance dynamics with application to noise and vibration control

Abstract

Estimation of models for both plant and disturbance dynamics is important in controller design applications which especially focus on the disturbance and vibration rejection. Several methods for low order model estimation on the basis of the closed-loop data exist in the literature, but fail to address the simultaneous estimation of low order models of both plant and disturbance dynamics. This dissertation contributes to the development of a new methodology to extend the results to low order disturbance model estimation, and apply these techniques to the control problems for disturbance rejection. In addition to the control relevant estimation problem, this dissertation also provides new tools for feedforward based disturbance rejection found in Active Noise Control (ANC) systems. We focus on the feedforward control algorithms that are one of the most popular methods to cancel low frequency sound where passive methods are ineffective. This dissertation shows that the feedforward filter design can also be seen as a model matching problem with the system model approximated on the basis of the expansion of orthonormal basis functions. Therefore, existing results on generalized FIR filters are exploited to provide feedforward compensation with the advantage of including the prior information of the system dynamics in the tapped delay line of the filter. It has the same linear parameter structure as FIR filter which is favorable for adaptation process. In the case that the acoustic coupling can not be neglected in the process of designing the feedforward filter, a dual-Youla parametrization is introduced and applied to estimate the possible perturbation of the feedforward filter and the robust stability of the closed loop system is enforced during the design of feedforward filter for active noise cancellation

Main Content
Current View