Hybrid Modeling of Dynamic Physical Systems:
An application in diagnosis
Abstract
Model-based monitoring and diagnosis of dynamic physical systems requires
well-contrained models of the system in an attempt to avoid computational
intractability. The key to modeling for diagnosis
is the ability to generate models that accurately describe both
steady state characteristics and the dynamics of system behavior.
The bond graph modeling formalism provides a suitable framework
that integrates and facilitates (1) generation of behavioral
constraints from physical laws, (2) expression of system
dynamics in terms of energy transfer between constituent elements, and
(3) modeling of steady state behavior as a special case of dynamic
behavior. In dynamic physical systems in general, and for the
purposes of diagnosis in particular, changes in the system may
require a number of model switches over time. Model switches take
place at discrete points in time, and combined with the typically
continuous characteristic of physical
systems, this naturally implies the need for hybrid models.
The discrete characteristics of the behavior are modeled as
finite state automata (FSA). The challenge in developing and
analyzing behavior with hybrid models is to integrate the FSA and
bond graph models. The end result of a set of discontinuous state
changes governed by the FSA always terminates in a valid continuous
state that can be analyzed using bond graph models. This research
develops a comprehensive hybrid modeling theory based on physical
principles which establishes (1) a formal hybrid
modeling approach, (2) an analysis method for model verification,
and (3) a physically correct behavior generation algorithm.
An application of hybrid models to diagnosis of complex dynamic
systems is also being developed.
A zipped up postscript file of a draft of my
area paper is 620159 bytes.
Committee
JANOS SZTIPANOVITS Professor of Electrical and Computer
Engineering
213 Jacobs Hall, Box 6306 Station B,
Nashville, TN 37235
(615) 322 3455
sztipaj@vuse.vanderbilt.edu
GAUTAM BISWAS Associate Professor of Computer Science
Vanderbilt University, 455 The Village at Vanderbilt, Box
1679 Station B,
Nashville, TN 37235
(615) 343 6204
biswas@vuse.vanderbilt.edu
GEORGE E. COOK Professor of Electrical and Computer Engineering
Vanderbilt University, 401b The Village at Vanderbilt, Box 1826 Station B,
Nashville, TN 37235
(615) 343 5032
cookge@vuse.vanderbilt.edu
GABOR KARSAI Assistant Professor of Electrical and Computer
Engineering
233 Jacobs Hall, Box 1824 Station B,
Nashville, TN 37235
(615) 322 2338
gabor@vuse.vanderbilt.edu
KENNETH A. DEBELAK Associate Professor of Chemical Engineering
307 Olin Hall, Box 1700 Station B, Nashville, TN 37235
(615) 322 2088
debelak@vuse.vanderbilt.edu
MICHAEL GOLDFARB Assistant Professor of Mechanical Engineering
505A Olin Hall, Box 1592 Station B, Nashville, TN 37235
(615) 343 6924
goldfarb@vuse.vanderbilt.edu
AMIT MISRA
misra@vuse.vanderbilt.edu