Predictive-control system application

PREDICTIVE control is used in this report to describe a form of automatic control in which the manipulated variable operating the controlled system is actuated by an estimate of the error which will exist at some future time. Repeated estimations of the future error are obtained by predicting ahead, on a fast-time base, both the reference and the controlled variable as well as some of their lower-order derivatives. Data smoothing and prediction of the reference may be required, and use of a model of the controlled system is needed. By virtue of being able to use fast switching at high energy levels and continually predicting a new value of future error, it is possible to obtain some of the desirable features of the bang-bang “optimum“ control as well as the dynamic programming concepts.