KEYWORDS: Sensors, Manufacturing, Process control, Photodetectors, Composite resins, Machine learning, Composites, Optical sensors, Data modeling, Control systems
This study introduces an innovative approach to enhance the utilization of carbon fiber thermosetting composites in advanced structural engineering by addressing the challenges of high manufacturing costs and limited production rates. We develop, deploy and test an ML pipeline utilizing PIC-based sensors (SOI technology, 220 nm thick, fabricated at IMEC’s MPW). They are based on a Bragg structure, packaged using ball lenses and suitable for operating at 180 degrees Celsius and 5 bar pressure. The focus is on accurately predicting two crucial parameters: Cure time and Temperature Overshoot, vital for determining the process duration and part quality. Using advanced tools and sensors, this study achieves a high prediction accuracy of 98% in millisecond scale while effectively handling the outliers. The ML pipeline allows the real-time process optimization of manufacturing process, minimizing the cost, and providing insights into the quality of the composite part through the in-depth monitoring of the process.
Fibre thermosetting composites play a major role in the engineering of advanced structures due to their combination of light weight and high strength and stiffness as well as the design flexibility. The high manufacturing cost and the inherently low production rates are the main limiting factors in increasing adoption of composites which can be overcome through the development of manufacturing strategies, materials and methodologies of process optimization and control. An accurate estimation of the stage of cure of thermosetting composites production is critical to deduce the overall process duration and ultimately the manufacturing costs. Challenges arise due to temperature overshoots and lack of direct measurement and control of the cure stage, particularly in thick components where the effects of the exothermic nature of the curing reaction and composite low thermal conductivity are more pronounced. To address these challenges and enabling the real-time process optimization, this study proposes a novel approach based on a machine learning (ML) model using simulation Finite Element Method (FEM) data as well as PIC-based photonic sensors realized on Silicon-on-Insulator (SOI) platform. Two robust Voting regressors, XGBoost and Light Gradient Boosting Machine, are used in the model to accurately (98% accuracy) predict two critical parameters: Cure time and Temperature Overshoot. Using photonic sensors to monitor the process in real time, we present experimental validation of Overshoot on manufacture RTM-6 aerospace composite parts.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.