Data-Driven Model Predictive Techniques for Unknown Linear Time Invariant Systems

This letter presents innovative data-driven model predictive control methods for unknown linear time-invariant systems. Two distinct approaches, both utilizing only input and state measurement data from the controlled system for controller design, are introduced. In both methods, a state feedback control law is devised at each time step, minimizing an infinite horizon objective function. The methods utilize the collected data to stabilize the unknown system. Finally, two examples are presented to validate the findings. Comparisons with existing research substantiate the effectiveness of these innovative approaches.