ES-MRAC: A new paradigm for adaptive control, theory and applications
Abstract
In this dissertation, we develop a method for the model reference adaptive control (MRAC) of systems via Extremum Seeking (ES), where extremum seeking is used as a means of estimating parameters in adaptive control. This provides a new approach to adaptive control that unifies real time optimization with adaptive control, thereby giving the designer far greater freedom in the choice of both cost functions and controller structure. Furthermore, conditions for persistency of excitation (PE) are explicit in our adaptive system. This renders the possibility of providing explicit and predictable rates for convergence of parameters, when the PE conditions are satisfied. Finally, our method provides a unified approach to adaptation that can be applied to many linear and nonlinear system structures, without having to modify the adaptation scheme. We show that our control/adaptation framework is extendable to a large class of nonlinear systems including adaptive backstepping and feedback linearization. As applications of our method, we apply it to the control of a two input, two output nonlinear model of a hypersonic vehicle in longitudinal plane, and the control of a rotary hydraulic crane with an underactuated mode. This dissertation opens up many interesting theoretical and practical problems to be addressed. A study on the noise rejection properties, the use of damped sinusoidal perturbations and stochastic perturbations in ES loops are just a few to name.
Degree
Ph.D.
Advisors
Ariyur, Purdue University.
Subject Area
Mechanical engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.