| Abstract Scope |
The RoboMet.3D, a semi-automated mechanical serial sectioning (MSS) tool, enables the acquisition of microstructural data across macroscopic length scales by layer-wise material removal and imaging, generating a 3D structure representation of complex multi-material components. This process unveils features and failure mechanisms hidden from traditional 2D and non-destructive methods. However, achieving uniform material removal is hindered by system inconsistencies, often demanding manual adjustments by skilled operators, thus extending data collection and necessitating extensive post-processing. We introduce a novel one-step model predictive control (MPC) framework integrated with a run-to-run (R2R) controller, designed to automate parameter adjustments, enhancing material removal precision through iterative feedback and disturbance management. This data-driven R2R-MPC controller ensures consistent removal rates, adapting seamlessly to varying material properties. Its superiority is validated against traditional methods through both simulations and empirical results, showcasing significant advancements in efficiency and reliability for 3D microstructural analysis. |