Date of Award
Fall 1998
Document Type
Dissertation - Restricted
Degree Name
Doctor of Philosophy (PhD)
Department
Electrical and Computer Engineering
First Advisor
Belfore, Lee A.
Second Advisor
Heinen, James A.
Third Advisor
Corliss, George F.
Abstract
With advancing technology, automated diagnosis of nonlinear systems is receiving considerable attention. The need for diagnostic tools exists in many different areas (medical measurements such as ECGs and industrial systems). The increasingly complex and highly nonlinear nature of these systems makes the research on diagnostic tools more challenging. Hence, automated diagnosis is an area of growth in which more and more researchers from many different branches of science conduct research in an attempt to obtain more dependable results. Research has been and is being conducted on diagnostic tools designed with different approaches such as expert systems, artificial neural networks, fuzzy logic, and probabilistic and attributed automata, singly or in combination. Expert systems use a knowledge-based approach. Diagnostic tools designed with other artificial intelligence schemes use a set of signals generated by the target system. The latter diagnostic tools are based on two main approaches: the decision-theoretic approach and, recently, the syntactic approach. The decision-theoretic approach focuses on the features of signals generated by the system. The syntactic approach examines the syntax of the structures in the signal to achieve a diagnosis. Diagnostic tools are designed to achieve a diagnosis on a certain system, and their performance is evaluated by the diagnostic accuracy they provide. The purpose of this study is to show that a diagnostic tool can be designed that is applicable to a variety of systems, and, yet, maintains diagnostic accuracy in the face of changing applications. In this work, we propose a new fuzzy syntactic approach to automated diagnosis. This approach combines the decision-theoretic and the syntactic approach. Time-sampled signals generated by the target system are transformed into a sequence of templates. Then, the template sequence undergoes a syntax analysis. The syntactic analysis is achieved using fuzziness which adds flexibility to the syntactic approach to handle noisy and imperfect information.