Multi-model DAE Based Observer and Filter Design

North Carolina State University, Raleigh NC

Investigators

Abstract

A different algebraic equation (DAE) is an implicit differential equation relating various quantities, their rates of changes, and control and source terms. DAEs are the natural way in which to describe many industrial and scientific processes. A DAE formulation often provides the quicker model development and more accurate modeling than when using explicit equations. However, using the DAE formulation on complex problems poses a number of technical challenges. Recently, it has been shown that a number of fundamental problems in control and systems theory including observer design, model identification, and failure detection can be naturally viewed as problems involving DAEs. In the past this has mainly been examined for lower dimensional linear systems. It is proposed to examine these problems for large complex nonlinear systems. In particular, systems with delays and mixed discrete continuous dynamics will be examined. Both theory and practical, robust, numerical algorithms will be developed. Particular questions to be examined include model identification, failure detection, observer design, system simulation, and fault isolation. The results will be applied to applications from the aerospace, chemical processing, and other industries.