Stability and Control of Singularly Perturbed Systems. Second Annual Report

摘要

This report surveys the results obtained during the second year of the three-year project entitled, Multimodel Strategies for Stochastic Models, and supported by the US Department of Energy, Electric Energy Systems Division. The report is divided into four parts. Part One presents a new method for studying stability of singularly perturbed systems using quadratic-type Lyapunov functions. The method is less conservative than the previous methods reported in the literature. Several examples are included to demonstrate that. Part Two is a continuation of our effort to study sampled-data control of singularly perturbed systems (see Part Three of the first annual report). Here we study singularly perturbed difference equations resulting from discretizing continious-time singularly perturbed systems. Part Three presents an important development in linear-quadratic Gaussian estimation and control of singularly perturbed systems. Near-optimum estimators and regulators are obtained using a special transformation to decouple slow and fast dynamics. Part Four considers a class of nonlinear singularly perturbed systems driven by wide-band noise and derives a reduced-order diffusion model that represents the behavior of the slow variables. (ERA citation 10:003020)