Information retrieval for intelligent information integration in nuclear power plant

The present study aimed at investigating functional requirements for information retrieval method to develop integrated information management for nuclear power plant and also aimed at developing a prototype system based on the extracted requirements. Among several requirements for supporting intelligent activities of plant operators, the focus has been set on the support function for effective retrieval from stored database, which is consistent with the cognitive process of operators. The functions of the quick wave form search based on partial similarity and of the data search based on hierarchically defined attributes have been realized with the man-machine interface for practical use. As for the quick wave form search based on partial similarity, the validity of this method has been evaluated through the numerical experiments for searching artificially induced wave form simulating signal anomalies. As for the data search based on hierarchically defined attributes, the validity of the proposed system has been shown through the process of possible user and system interaction for anomaly detection and retrospective analysis when some anomalies arise in terms of frequency characteristics in a specific frequency band.