Removal Thickness Prediction and Process‐Window Identification in Laser Cleaning of Multilayer Aerospace Coatings

ABSTRACT

Precise control of removal thickness is critical for laser cleaning of multilayer aerospace coatings particularly for selective removal without substrate exposure. However, the process response becomes highly sensitive near coating interfaces, making reliable parameter selection difficult. In this study, a regression‐based model was developed to predict removal thickness and identify process windows for controlled multilayer laser cleaning. Experiments were conducted on coatings deposited on 2A12 aluminum alloy, and 200 samples were obtained using a pulsed fiber laser. The model achieved high predictive accuracy with an R ² of 0.970, a root mean squared error (RMSE) of 1.978 μm, and a mean absolute error (MAE) of 1.506 μm. The results reveal distinct nonlinear behavior across the parameter space with narrow transition regions governing process sensitivity. Based on experimentally calibrated boundaries, a process window was identified and validated, providing practical guidance for controlled coating removal.