Intelligent control in safety systems: criteria for acceptance in the nuclear power industry

Abstract

Hierarchical systems incorporating intelligent controllers have been proposed for use in process control of complex systems, including nuclear power plants. It is noted that, in contrast to these proposals, the controllers that are used in today's operational plants are very simple, and are likely to remain so unless the industry can be convinced that more sophisticated control concepts are compatible with its demanding requirements for reliability, validation, and verification. The authors analyze these requirements and propose a concept of system architecture that has the capacity to meet them. Essential properties of the proposed methodology are modularity of the design and distributed autonomy within each module. These features are needed to ensure a system that can be implemented and tested incrementally and that will exhibit robustness in the event of plant or control system failures that render individual controller subsystems ineffective. The design of an intelligent controller for a feedwater heater provides an illustrative example of the proposed concept. >