Data-based tuning strategies for positioning system in high-speed high-precision motion control

Abstract

This Ph.D. study explores a series of novel data-based tuning methods to find optimal parameters for feedforward controller, feedback controller and disturbance observer, in high-speed high-precision motion control. This aims to improve tracking performance of positioning system utilizing batch-to-batch real-time data. The key research content and work are therefore summarized as follows:
Development of a data-based tuning method which tune the parameters of two inverse models used in feedforward controller and disturbance observer simultaneously. This method significantly improve the tracking performance of positioning system without increasing the tuning complexity.
Development of data-based tuning method on bumpless feedforward controller, to improve servo performance of wire-bonding machine when implementing multi-phase trajectory. The bumpless feedforward controller in which multiple sub-controller switched consecutively, is tuned by the specified data-based tuning method under the guidance of explicitly derived gradient expressions. The motion phases of the multi-phase trajectories are cope with the tuned sub-controllers. Therefore, the system’s inverse dynamics is captured by the bumpless feedforward controller more accurately, and the tracking performance of the positioning stage of wirebonding machine is enhanced.
Development of data-based tuning method on bumpless inverse model used in both feedforward controller and disturbance observer. The bumpless switching mechanism is introduced into disturbance observer at first. Then, the associated data-based tuning method is developed to tune both bumpless inverse model in feedforward controller and disturbance observer. Besides, the perturbation experiment is revised to reduce the experimental burden throughout the tuning process. With tuned bumpless inverse model, the tracking performance of positioning system is improved.
Development of data-based bumpless tuning method on multiple-input-multiple-output (MIMO) system. The bumpless feedforward controller associated with the data-based tuning approach is extended into MIMO system. The gradient expressions is explicitly derived with respect to time interval, degree-of-freedom of the system, and is proven to be a tensor. A data representative of the MIMO sensitivity function is constructed using step-response-based perturbation, helping to reduce experimental burden. The enhanced performance of it is experimentally verified in the positioning stage of wire-bonding machines.
In summary, with the switching controller tuned via data-based strategies, the control algorithm achieves better tracking performance while maintaining some robustness against trajectory variations. This Ph.D. work investigates the data-based tuning approaches in high-speed high-precision motion control. By developing the tuning methods on the switching controller with bumpless transfer techniques, the tracking performance of positioning system is enhanced by exploiting information from real-time operational data.

Date of Award	15 Mar 2026
Original language	English
Awarding Institution	University of Nottingham
Supervisor	Dunant Halim (Supervisor), Silu Chen (Supervisor) & John Xu (Supervisor)