Multiple Model Adaptive Regulation

A single controller may be inadequate for systems that experience structural changes that arise, for example, from component failures. Such systems are often modeled by a family of plants with structural diversity. At any given time the appropriate plant model is uncertain. Adaptation is required for the parameter dependent family of plants but continuous adaptive regulation is limited by relative degree and right half plane zeros. A form of adaptive regulation is presented that accommodates these changes. The Multiple Model Adaptive Regulator selects a controller from a predefined set to achieve performance goals. In general, the set of controllers is finite although the family of plants may be continuous. The set of controllers accommodates a structurally diverse family of plants. A multiple model controller design has two subproblems, covering and switching. The covering subproblem is to design a small set of controllers such that each possible plant is stabilized by at least one controller. The switching subproblem is to select a stabilizing controller from the set of controllers. In this research, the covering and switching subproblems are solved with LQR state feedback and Lyapunov function switch logic respectively. The LQR and Common Quadratic Lyapunov Function design problems are combined into a set of Linear Matrix Inequalities (LMI) and concurrently solved. Design constraints on the open loop plant and the regulator are presented. The multiple model adaptive regulator is applied to systems with diverse zero structure.Ph.D., Mechanical Engineering -- Drexel University, 201