A sequence of simulation and physical system testbed facilities for developing, verifying, and validating intelligent control for commercial nuclear power plants is proposed. The experimental development sequence starts with the theoretical conceptualization and simulation of a candidate control law. The second step, which focuses this project to the nuclear aspects of power system, formulates the candidate controller for demonstration on an ultra-safe university research reactor. The third step of the sequence formulates the controller for demonstration on the passively safe Experimental Breeder Reactor (EBR- II) power plant. The 20 MW(e) EBR- II contains a Liquid Metal-cooled Reactor (LMR) with demonstrated passive safety features which make it especially well-suited for control experiments. The goal of the experimental sequence is to safely and economically identify the best intelligent control concepts for commercial application. The experimental development sequence will be created as a general purpose long- term capability for reducing theoretical advances in intelligent control to practice. To efficiently establish the sequence, our recent research to improve reactor temperature response using optimal, robust, neural network and fuzzy logic control will be incorporated in a reconfigurable controller architecture which is initially governed by a learning systems-based intelligent supervisor.
