A scheduling quasi-minmax MPC for LPV systems

Presents a scheduling model predictive controller for constrained discrete linear parameter variable (LPV) systems. At each sampling time the parameters of the LPV plant are measured and their future values are assumed to vary arbitrarily inside a specified polytope. Given this information along with the current state the algorithm minimizes online an upper bound on the quasi-worst case infinite horizon objective function subject to input and output constraints. The first computed input is implemented and the calculations are repeated at the next sampling time. Optimization can be solved by semi-definite programming including constraints that are expressed as LMIs (linear matrix inequalities) and is convex. The resulting receding horizon controller guarantees stability if the optimization has a feasible solution