Abstract:
This paper considers the problem of fault diagnosis in critical-purpose discrete-event systems described by the fuzzy finite state automaton (FSA) model. A solution method involving the mathematical apparatus of fuzzy logic is proposed. Fuzzy logic operations are described, and the concept of the determinizer of a fuzzy FSA is introduced. A diagnosis scheme that forms a structured residual vector is given. This scheme contains several channels (according to the number of possible faults in the system). Each channel is based on an observer, i.e., a determinizer of a special fuzzy FSA that simultaneously considers the possibility of both correct and incorrect transitions of the automaton (the normal operation of the system and the occurrence of a system fault, respectively). Another part of the channel is the decision block. Some ways to design the observer and the decision block are proposed. The features of the solution method are illustrated on the example of error monitoring for human operators in IT systems.
Keywords:discrete-event systems, fuzzy logic, fuzzy finite state automata, determinizer, fault diagnosis, IT systems, monitoring.
Fulltext:
PDF file (2068 kB)