Abstract


Online tasks of diagnosing disturbances and planning remedial actions at the human-machine interface of Nuclear Power Plant (NPP) are difficult work for human operators. To reduce the workload of operators, Computerized Operator Support System (COSS) has been studied to model the causality and find the root cause of a failure situation of NPP by semiotic analysis. In this paper, a graphical method of semiotic analysis called Multilevel Flow Models (MFM) is introduced to augment the existing post-trip monitor system based on neural network for alarm analysis of NPP. A typical 4-loop Pressurized Water Reactor (PWR) is employed as the reference system and various accident simulations have been conducted by using RELAP5/MOD2. A desk-top examination is made to confirm the validity of the proposed method.

Powered by JDBS Ver.2.1.0
(C) 2020 Hirotake Ishii