Gaussian function neural network-based self-tuning control of KMRR

This paper describes a self-tuning control of a nuclear reactor system for which no model exists, and for which the only available data are a set of input-output measurements. The use of artificial neural network in nonlinear model-based adaptive control, both as a plant model and a controller, is investigated. A neural network called Gaussian function network (GFN) is used for the one-step-ahead predictive control to track the desired plant output. The effectiveness of the controller is demonstrated by application of the method to the power tracking control of Korea Multipurpose Research Reactor (KMRR). (Author)