Infeasibility handling in linear MPC subject to prioritized constraints

All practical MPC implementations should have a means to recover from infeasibility. We propose an algorithm designed for linear state-space MPC which transforms an infeasible MPC optimization problem into a feasible one. The algorithms handles possible prioritizations among the constraints explicitly. Prioritized constraints can be seen as an intuitive and structural way to impose process knowledge and control objectives on the controlled process. The algorithm minimizes the constraint violations by solving a series of optimization problems, and the violation of a given constraint is minimized without affecting the higher prioritized constraints. An example shows the effect of implementing this algorithm on a simple process.