Adaptive model predictive control with robust constraint satisfaction

Adaptive control for constrained, linear systems is addressed and a solution based on Model Predictive Control (MPC) and set-membership system identification is presented. The paper introduces a computationally tractable solution which uses observations of past state and input trajectories to update the model and improve control performance while maintaining guaranteed constraint satisfaction and recursive feasibility. The developed approach is applied to a stabilizing MPC scheme and practical stability under persistent, additive disturbance is proved. A numerical example and brief comparison with non-adaptive MPC is provided