Analysis of optimal composite feedback‐feedforward control

We have developed analytic design methods for combination feedback-feedforward control systems and have evaluated systems yielding optimal performance while subject to constraints commonly encountered in the chemical industry. Using the mathematical techniques pioneered by Wiener for the solution of the design equations, we have based the optimization on minimization of the mean square output of a system subject to a random disturbance. Side conditions for the constraint on mean square control effort, signal-to-noise ratio in the feedback system, and minimization of error output caused by misidentification of plant parameters were found to be necessary to give physically realizable and meaningful designs. The analytic design methods are useful for analysis of control system performance and capabilities under a variety of constraints, but the optimal designs are marginally superior to ideal or invariant feedforward controllers coupled with tuned proportional feedback controls. The principal improvement in the optimal design is in conservation of control effort when compromises in system performance are necessary because of this restriction.