Automated fault-tree generation for operational fault diagnosis

This paper describes an expert system approach to offline generation and optimization of a fault-tree for the online diagnosis of a modern train, incorporating the knowledge and experience of the train manufacturer and the train operators. The size of the problem is such that explicit formulation of the fault-tree is very complicated. The knowledge however, is implicitly available in the description of the faults in terms of symptom codes, answers to additional questions and repair actions. Case-based reasoning is selected for the implementation to facilitate the automatic generation, consistency checking and maintenance of the fault-tree. Special attention is given to the processing speed needed for online use on the train