A Technique for On-Line Steady-State Optimizing Control

A technique for on-line steady-state optimizing control is developed for systems subject to disturbance inputs. The parameters of the control algorithm are updated so as to maximize the expected system performance, based on a prediction of the frequency distribution of the disturbance inputs over the next time interval of operation. The statistical parameters of the distribution are assumed to be slowly time-varying and the adaptation procedure includes their updating, based on past observations of the disturbance variables. Some of the advantages and limitations of the technique are demonstrated by application to the control of a simulated stirred-tank reactor.