Artificial neural network-based modeling of sustainable plasma process parameters for predicting wettability of aircraft composite surfaces

Abstract

Modeling the wettability behavior of Carbon Fiber Reinforced Polymer (CFRP) materials treated with atmospheric pressure plasma is crucial for efficient and sustainable manufacturing processes, especially for aerospace applications. In this study, plasma process parameters were modeled and 3D-mapped in order to predict wettability of CFRP by using Artificial Neural Networks (ANN). Data from plasma treatment process parameters was used to train the ANN model, and its performance was evaluated using error analysis and correlation coefficients. Contact angle, surface roughness, and surface energy measurements were carried out to assess the wettability of the plasma-treated composites. Furthermore, X-ray Photoelectron Spectroscopy (XPS) and Atomic Force Microscopy (AFM) analyses confirmed the chemical changes and indicated that the surface topography was maintained. A clear consistency was observed between the experimental measurements and the model predictions. The determination coefficient (R2) of Water Contact Angle (WCA) was 0.9839 for the training and 0.9513 for the testing data. Similarly, for the Diiodomethane Contact Angle (DCA), the R2 values were 0.9837 and 0.9483, demonstrating reliable predictive capability. The model also revealed that a nozzle speed of 10 mm/ s and a distance of 11 mm produced optimal surface characteristics, whereas 30 mm/s and 20 mm corresponded to insufficient activation energy.