Potential application of hyperspectral imaging for monitoring resin cure in composite manufacturing via liquid resin infusion.

This study investigates the feasibility of using Near-Infrared Hyperspectral Imaging (HSI) for monitoring the curing state in Liquid Resin Infusion (LRI) manufacturing. By capturing spectral reflectance data throughout the heating and curing phases of the process, we explore the potential of HSI as a non-destructive technique for assessing resin polymerization. Compared to conventional sensor-based methods, HSI offers two key advantages: it eliminates the need to embed sensors within the manufactured part and provides spatially resolved information across the entire surface, enabling the detection of cure inhomogeneities. First, an in-depth study was conducted on the spectral features captured by NIR-HSI and their temporal evolution throughout the process, comparing them with key absorption bands related to the epoxy resin used in LRI. Afterwards, spectral data was correlated with the curing state inferred from commercial dielectric sensors using regression models, including Partial Least Squares Regression (PLS-R) and Support Vector Regression (SVR). These regression models were trained using the full spectral range of a specific spatial region of interest, as well as wavelength subsets selected through feature selection techniques such as Mutual Information (MI) and Random Forest Regressor (RF). The results indicate a strong correlation between the spectral data and the inferred curing state, with the SVR model trained on the full spectral range achieving the highest performance (R=0.9983, RMSE = 0.0114, MAE = 0.0086). Moreover, the predicted curing maps display consistent trends across all defined regions of interest (ROIs), further demonstrating the capability of HSI to capture spatial variations in the curing process. These findings reinforce the potential of Near-Infrared Hyperspectral Imaging as a powerful, non-destructive tool for real-time monitoring of resin polymerization in LRI manufacturing.