Development and implementation of a power system fault diagnosis expert system

This paper describes a fault diagnosis expert system installed at the Tohoku Electric Power Company. The main features of this system are careful selection of the inferencing input data, rapid inferencing, integration of the expert system with other systems in a practical structure, and the adoption of a domain shell. This system aims for improved practicability by using time-tagged data from circuit breakers, protective relays, and automatic reclosing relays in addition to the input data used in earlier systems. Furthermore, this system also uses data from fault detection systems that locate fault points within electric stations. This system uses an AI-specific back-end processor to perform inferencing rapidly. Additionally, this fault diagnosis expert system is interfaced and integrated with a restorative operations expert system, an intelligent alarm processing system, and a protective relay setting and management system. The authors developed and adopted a power system fault diagnosis domain shell to ease system development, and used the protective relay operation simulation function of a protective relay setting and management system for system verification. >