A design methodology for self-tuning control of systems with inherent conflicts

Abstract

Outlined in this thesis is a methodology for the design of self-tuning controllers for systems with inherent control conflicts. These conflicts arise from either physical phenomena or user specifications. An intrinsic part of this methodology is the extensive collection and computer-aided incorporation of a priori information in the design process to enable realization of controllers of relatively simple structure and of low on-line computational requirements. Such simple solutions to complex problems are often overlooked when a priori information is unnecessarily neglected in some of the more mathematically elegant self-tuning algorithms proposed to date. Another intrinsic element of the methodology is the use of integral manifold and averaging techniques to simplify the analysis and design of self-tuning control of real processes. This serves not to confuse the design engineer, but to demonstrate the practical nature of some of the control theorist's tools.