A Practically Stabilizing Model Predictive Control Scheme for Switched Affine Systems

This letter presents the design and analysis of a practically stabilizing model predictive control (MPC) scheme for a class of discrete-time switched affine systems with at least one stable subsystem. A pruning algorithm with cost approximation is proposed to ensure the practical stability of the MPC scheme while reducing the online computations. A criterion for selecting the pruning parameters to ensure the feasibility of the MPC scheme is presented, which is formulated as a convex optimization problem. With the proposed pruning, the value function of the MPC scheme becomes a practical-Lyapunov function which guarantees global practical stability of the switched affine system for arbitrary prediction horizons. The MPC scheme is applied to switching power converters and simulation results are given for a buck converter which is a DC-DC converter.