Bumpless Transfer Model Predictive Control for Markov Jump Linear Systems

This study is concerned with the bumpless transfer control (BTC) for discrete-time Markov jump linear systems (MJLSs) subject to constraints on states and inputs. A new approach called bumpless transfer model predictive control (BT-MPC) is proposed, which simultaneously optimizes the stabilization performance and bumpless transfer performance in a receding horizon manner. By a scenario-based control law, the fixed-duration transition progress is generalized to the mutable-duration one, so that the BTC scheme can be applied to MJLSs. Compared with existing BTC approaches, BT-MPC avoids the requirement of predesigned controller gains or gain variations, thereby reducing the conservatism and simplifying the parameter tuning. The recursive feasibility of BT-MPC and the mean-square stability of the underlying systems are theoretically guaranteed, and a simulation example is provided to validate the developed results.