Integration of Expert Systems and Neural Networks in Machine Diagnosis

摘要

Neural networks are effective at classifying acoustic or vibrational information which can augment machine diagnosis. Expert systems are effective at representing more explicit diagnostic knowledge gained from experts, as well as providing explanations of its diagnoses, and performing sequential actions such as test sequencing, repair, and repair validation. The feasibility of integrating expert system and neural network technologies has been demonstrated with the completion, during Phase I, of a prototype hybrid system. The system uses neural networks to perform an initial diagnosis via acoustic signal recognition, and it uses an expert system to perform follow-up tests leading to a specific diagnosis. This system successfully diagnoses a sample mechanical fault using acoustic information and expert-level knowledge.