Domain-specific design of a non-destructive health monitoring expert system

This paper introduces a signal-to-symbol transformation model and proposes a corresponding software architecture for implementing an event-synchronous flaw signal diagnostic mechanism in a health monitoring expert system. In the proposed concept, to filter out input signal data effectively, and to modularize the process of knowledge processing and elicitation, the task of inspection can be delegated to two knowledge spaces each of which has a proper knowledge processing scheme to match the properties of its own task. The task for representing signal-specific knowledge which detects the signal patterns (events) leading to any harmful flaw is implemented by integrating a symbolic representation and syntactic parsing concept based on fuzzy set theory. The task for domain-specific knowledge evaluating the characteristics of the events is built based upon a rule-based expert system concept on the top of the fuzzy symbolic processing architecture. To propose a guideline of system integration, the signal-specific knowledge and domain-specific knowledge are conceptually modeled using object-oriented abstraction hierarchy. The proposed architecture has been verified by implementing a prototype which was developed to automatically interpret non-destructive evaluation signals for inspecting health of tubes used in nuclear power plants.