Receding horizon H∞ predictive control for systems with input saturation

A receding horizon H, predictive control method is derived by solving a min-max problem in nonrecursive form. The min-max cost index is converted to a quadratic form which, for systems with input saturation, can be minimised using QP. Stability conditions and H, norm bounds on disturbance rejection are obtained. Without input saturation, stabilising terminal weights guaranteeing finite H, norm bounds always exist, but the guaranteed H, norm bound may be conservative; this is remedied through the use of closed-loop prediction. Feasible sets for state and disturbances are derived, for which stability can be guaranteed. The algorithm and weight selection procedures are given in terms of LMIs.