A Suboptimal On-Line Discrete Controller with Bounded Control Variables

A method to synthesize the control signal on-line, by using the form of an optimization problem, is examined. Control of linear processes is considered, and bounds or inequality constraints on the control variable are attacked directly, without assigning penalties on the performance criterion. The optimal-control problem is formulated as a nonlinear programming problem, and the solution is obtained through a co-ordinate-wise gradient method. Simulation of the on-line controller is performed; a marked improvement in accuracy over conventional sampled data systems is shown. The method, as presently employed, can consider bounds on the magnitude or on the rate of change of the control variable.