A Reconfigurable Control Strategy for Distributed Digital Process Control

A reconfigurable control scheme is proposed which, unlike a preprogrammed one, uses stochastic automata to learn the current operating status of the environment (i.e., the plant, controller, and communication network) by dynamically monitoring the system performance and then switching to the appropriate controller on the basis of these observations. The potential applicability of this reconfigurable control scheme to electric power plants is being investigated. The plant under consideration is the Experimental Breeder Reactor (EBR-II) at the Argonne National Laboratory site in Idaho. The distributed control system is emulated on a ring network where the individual subsystems are hosted as follows: (1) the reconfigurable control modules are located in one of the network modules called Multifunction Controller; (2) the learning modules are resident in a VAX 11/785 mainframe computer; and (3) a detailed model of the plant under control is executed in the same mainframe. This configuration is a true representation of the network-based control system in the sense that it operates in real time and is capable of interacting with the actual plant.