Data-Driven Dynamics Decomposition of Dual-Drive Gantry Stage and Robust Decoupling Control

Abstract

This article presents a frequency-domain control framework for dual-drive gantry stages based on data-driven mode decomposition. The system dynamics are measured and decoupled using tensor decomposition, and the resulting modes are reconfigured into physically meaningful task coordinates such as linear and rotational motion. This task-space representation facilitates intuitive analysis and enables the independent design of two-degree-of-freedom (TDOF) controllers for each mode. The proposed approach improves control bandwidth and robustness by minimizing dynamic coupling across a wide frequency range. Theoretical analysis and experimental validation confirm the following: 1) the data-driven decomposition significantly reduces coupling effects over the entire operating spectrum; 2) the task-oriented dynamic model provides a structured basis for model-based control; and 3) the resulting TDOF controller outperforms conventional approaches in terms of precision and robustness.