Stochastic optimal control of ARMAX model

Simultaneous state estimation and parameter tracking of time varying input-output ARMAX model with known noise parameters was described recently. Optimal control strategies of such model are developed in this contribution. Certainty equivalent and cautious LQ strategies (Linear system and Quadratic criterion) are obtained. Certainty equivalent strategies operate with state and parameter means only and so neglect the uncertainty while cautious strategies take parameter and state estimation uncertainty into the consideration.