Physics Journal of the Indonesian Physical Society
REACTOR MONITORING WITH NEURO-EXPERT SYSTEM FOR RSG-GAS
Surian Pinem1, Tukiran Surbakti1 and Kunihiko Nabeshima2
1Pusat Pengembangan Teknologi Reaktor Riset, BATAN, Serpong, Indonesia2Japan Atomic Energy Research Institute, Tokai-mura, Ibaraki-ken, 319-1195, Japan
This paper addresses the neuro-expert monitoring system applied to detect the reactor core. The neural network can be utilized to model the reactor dynamics with normal operation data and to detect a symptom of small anomalies earlier than the conventional alarm system. The expert system can diagnose the reactor status from the operational information and the output of the neural network. The off-line test results showed that the neural network successfully monitored the reactor status during steady state operation. The deviation between the measured and the estimated values is always in the error boundary, so that the monitoring system indicates the reactor status operated normally.