Modelling of the knowledge for monitoring expert systems in nuclear power plant safety

Safety operation support systems for NPP faced problems of difficult solutions along their development. This work presents possible solution to such problems, and contribute to enhance the reliability and performance of such system using Artificial Intelligence. Knowledge representation is capital in this work since it express the dependence on variables in a rather natural way. therefore, it makes intrinsic the concepts of synchronism and concurrence in real-time approach. Other advantages are easy V and V processes and simplification of the system maintenance procedures. The inference process is carried out through the rules that are generated from knowledge base. These rules are charged following a conflict resolution optimized for time-real approach. The real application used to validate the model efficiency, consists in part of SICA (Integrated System of the Angra-1 Computers). The application results revealed very positive reducing the quantity of the SICA conventional software code programming. As far the system performance. the knowledge structures and the conflict resolution strategy adopted allowed for guarantee not only the time control for inference, but also a response time compatible with that requested for power plant safety support. (author) 12 refs., 4 figs