A class of peak regulators with fixed feedback control laws

This study concerns control laws which are optimal in the sense of minimizing a class of performance indices which penalize peak trajectory error. For the time-invariant case with infinite control horizon, the optimal control is shown to exist, be unique, and be a time-invariant feed-back map requiring current state and worst-error-to-date. This law is asymptotically stable if and only if the (linear) model has no J?? axis poles. A linear terminal control law is explicitly calculated, and necessary and sufficient conditons which characterize all optimal solutions presented.