Disturbance Observer-Based Data Driven Model Predictive Tracking Control of Linear Systems

This paper introduces a new data-driven MPC structure based on two offline and online parts to achieve the robust and constrained performance in an optimal scheme. In the first step, according to the model matching condition, an offline data-driven controller is designed to reach the tracking performance. In addition, to reduce the effects of the external disturbance, a data-driven-based disturbance observer is presented to estimate the external disturbance. Therefore, the robustness against the external disturbances is achieved in an offline procedure. Then, a data-driven model predictive control (MPC) is structured based on a data-driven-based model of a stabilized system. In other words, the overall controller is configured such that the limitations of the system states and control input are considered in the control design process. Moreover, by employing the move blocking strategy, the online computational burden of the suggested controller is greatly reduced. To further improve of the feasibility problem, an ellipsoidal terminal (ET) constraint is considered. The rows number of the blocking matrix influences on the ET set which leads to feasibility enhancement. So, the main contributions of the presented data-driven controller are feasibility improvement and reducing online computational burden in an optimal and constrained scheme which are illustrated in the simulation section.