A comprehensive fault diagnostic system using artificial intelligence for sub-transmission and urban distribution networks

This paper describes an intelligent diagnostic system for an interconnected distribution network developed to assist the system operator with fault identification during restoration. The intelligent process utilizes only those data available in a standard SCADA system such as the post fault network status, the list of the tripped breakers, main protection alarm, and the conventional event log. The fault diagnostic system is implemented by three independent mechanisms, namely the generic core rule, the generic relay setting inference and the specific post-fault network matching and learning. The generic core rule generates various possible fault locations and the generic relay inference examines whether each possible fault location is logical and valid. The specific network matching compares whether the post fault network and the related tripped breakers are identical to a previous fault event. Test results obtained from two distribution networks confirm that the developed system is practical, reliable and accurate.