| The objective of this research is to develop an integrated machine fault diagnosis system. Both rule-based and model-based reasoning are utilized to ensure the functionality of diagnosis. This system is composed of five modules. They are: (1) machine description module, (2) diagnosis hierarchy module, (3) integrated inference module, (4) knowledge compilation and refinement module, and (5) interface module. The machine description module allows the modeling of physical mechanisms and their specific diagnosis details. The diagnosis hierarchy module captures the expertise of knowledge engineers in constructing a rule base in hierarchical levels of diagnoses. The integrated inference module integrates rule-based diagnosis and model-based diagnosis into a hybrid diagnosis system. The knowledge compilation and refinement module generates an explanation structure from causal knowledge and cases, them into heuristic rules, and refine the existing rule base. The interface module provides the necessary generalized user interface features for all applications in the system. These five modules form a framework of a customized machine fault diagnosis shell environment. The computer system has been intensively validated and the validation results are satisfactory. |
