An Industrial Application of a Self-Tuning Regulator

An implementation of a self-tuning regulator is described. The regulator algorithm consists of two steps. Firstly the parameters of a model of the process are estimated in real time using the method of least squares. Secondly a minimum variance regulator based on the estimated parameters is computed at each sampling interval. Implementation of the algorithm on a process computer is discussed. Questions concerning program structure and storage requirements are considered. Results from the moisture content control on a paper machine are presented. The experiences are very good concerning the start-up behaviour as well as the stationary behaviour of the self-tuning regulator.