Data-based tracking control for nonminimum-phase systems

Proposes a tracking control design for exponentially stable linear time invariant systems. Unlike most conventional designs, the authors´ design dispenses with knowledge of the system parameters. The only information required is a set of system input and output data, which is obtained through a series of testing experiments. No parameter identification process is introduced explicitly or implicitly. Instead, the control input is synthesized directly based on the obtained system data such that the more data is utilized, the smaller L₂-norm of the tracking error is achieved. An important contribution of the proposed design is that it can be applied to the tracking control of nonminimum-phase systems successfully for either finite or infinite operation time. In the infinite operation time case the proposed control is shown to be a discretized state feedback control, and a minimum preview time of the reference trajectory is imposed by the stability concern of the closed-loop system. A method for calculating the minimum preview time from the experimental system data is also provided. Simulation studies show that for a nonminimum-phase system the closer the dominant unstable zeros to the imaginary axis, the longer the minimum preview time required