A data-driven controller for position tracking of a long-stroke piezoelectric actuator

Hysteresis and other nonlinearities exhibited by piezoelectric materials significantly limit the tracking capability of piezoelectric actuators (PAs). To eliminate these effects, several model-based controllers adopted the inverse model-based approach. However, these controllers are overly sensitive to the uncertainties in the model parameters and can lead to poor tracking performance. This paper proposes a novel model-free learning-based controller to achieve high position tracking performance. The new combined feedforward-feedback controller synthesizes the control signal using the input and output data. The paper uses the passivity theory and the Lyapunov method to prove the system closed-loop stability and the boundedness of the tracking errors and the control signals. It further develops "intelligent" rules to ensure the system's robustness to transient disturbances. The proposed controller was tested using a commercial long-stroke piezoelectric actuator to assess its tracking performance.