Inversion-based feedforward-feedback control: theory and implementation to high-speed atomic force microscope imaging

In this dissertation, a suite of inversion-based feedforward-feedback control techniques are developed and applied to achieve high speed AFM imaging. In the last decade, great efforts have been made in developing the inversion-based feedforward control as an effective approach for precision output tracking. Such efforts are facilitated by the fruitful results obtained in the stable-inversion theory, including, mainly, the bounded inverse of nonminimum-phase systems, the preview-based inversion method that quantified the effect of the future desired trajectory on the inverse input, the consideration of the model uncertainties in the system inverse, and the integration of inversion with feedback and iterative control. However, challenges still exist in those inversion-based approaches. For example, although it has been shown that the inversion-based iterative control (IIC) technique can effectively compensate for the vibrational dynamics during the output tracking in the repetitive applications, however, compensatingfor both the hysteresis effect and the dynamics effect simultaneously using the IIC approach has not been established yet. Moreover, the current design of the inversion-based feedforwardfeedback two-degree-of-freedom (2DOF) controller is ad-hoc, and the minimization of the model uncertainty effects on the feedforward control has not been addressed. Furthermore, although it is possible to combine system inversion with both iterative learning and feedback control in the so-called current cycle feedback iterative learning control (CCF-ILC) approach, the current controller design is limited to be casual and the use of such CCF-ILC approach for rejecting slowly varying periodic disturbance has not been explored. These challenges, as magnified in applications such as high-speed AFM imaging, motivate the research of this dissertation. Particularly, it is shown that the IIC approach can effectively compensate for both the hysteresis and vibrational dynamics effects of smart actuators. The convergence of