Applicability of autoregressive modeling technique to plant monitoring system of fast breeder reactors

In this report, the results of examination and evaluation on the effectiveness and problems of self regression model method are summarized, as this method is expective to be one of the most powerful methods for examining the operation monitoring system for fast reactors. In this analysis, the data of M series turbulence application experiment and the data of steady reactor noise obtained from the fast experimental reactor ''Joyo'' were used, and emphasis was placed on the applicability to actual plants. It is the central techniques of this monitoring system to make the behavior forecast with models for respective subprocesses into which a plant was divided, and to detect the occurrence of abnormality from the increase of difference between the forecast and measured values. In this case, the accuracy of models for normal operation is the most decisive factor, therefore, its evaluation was performed preponderantly. Also, the possibility of extracting information on plant dynamic characteristics from the self regression model and identifying the cause of detected abnormality was investigated. The successful results were obtained. (Kako, I.)