Electrical fault diagnosis using expert systems

A comprehensive approach to the development and organization of an expert system designed to assist electricians in diagnosing electrical faults in mining machinery is discussed. The system includes diagnostic, reference, and tutorial programs. The diagnostic portion of the system is based on a causal-reasoning approach, which is used to establish a hierarchy of symptoms and the subsequent production rules for identifying component-level failures in the motor-control circuit of a shuttle car. With this approach, conclusions are developed directly from knowledge of the circuit and its behavior (in both working and failure modes) based on first principles, rather than the heuristics of a domain expert. The symptoms and their hierarchal structure are formulated by creating "what if' scenarios and performing a qualitative circuit analysis for each fault. Techniques for incorporating interactive graphical displays of the circuit schematic into the diagnostic portion of the system are also presented. Supportive computer programs that provide reference material and tutorials on the circuit are also described.