Predictive switching supervisory control of persistently disturbed input-saturated plants (Q705436)

The paper presents a solution to the regulation problem for input-saturated linear time-invariant plants described by discrete-time state-space models. The plants are possibly subject to persistent bounded disturbances of unknown arbitrary magnitude. At any time, the state feedback of the plant is ensured by a gain matrix selected from a finite family of pre-designed candidates. Each candidate gain matrix is tuned on to a different control-horizon, according to a predictive control strategy. A supervisory switching control scheme selects the appropriate gain matrix, relying on the information about the previous gain matrix and the current state of the plant. It is proved that the control scheme guarantees global asymptotic and semi-global exponential stability in the ideal noiseless case. The approach to the noisy case considers stability and boundedness under arbitrary \(\ell_\infty\) disturbances. Two simulation examples are also reported.