A case learning tool for operations of power systems

This paper presents the development of a case learning tool for operation support (CALTOS). The authors have proposed a new framework of automatic knowledge acquisition through machine learning. The learning algorithm named HCL produces operational rules from cases in the past. The HCL has been built into CALTOS and it provides an environment for rapid and efficient construction of intelligent systems which can update its knowledge base without explicit revision. This paper argues the difficulty of intelligent systems development and proposes a solution with CALTOS. Here, two developed systems are introduced, one is for an extra-high voltage substation and the other is for a dispatching office. A dispatching office generates instruction sequences for its supervised substations or related offices, while a substation produces switching sequences of devices for each instruction given by its supervising office. These systems have been developed just by means of adding facility data and operational cases of the target substation or the dispatching office.