DEVELOPMENT AND PERFORMANCE OF A NEURAL-NETWORK PREDICTIVE CONTROLLER

Neural-network techniques are investigated in an application to the id entification and subsequent on-line control of a process exhibiting no n-linearities and typical disturbances. The design and development of a neural-network process model from measured data is described, and pr actical aspects of the identification procedure are discussed. Results demonstrate that the developed neural-network representation of the p rocess dynamics is sufficiently accurate to be used independently from the process, emulating the process response from only process input i nformation. Accurate long-range predictions from the neural-network mo del are mainly due to the use of a novel spread encoding technique for representing data in the network. Implementation of a predictive cont rol strategy incorporating the identified neural-network model is desc ribed, and on-line results illustrate the improvements in control perf ormance that can be achieved when compared to conventional proportiona l-plus-integral control.