Curing simulation and data-driven curing curve prediction of thermoset composites.

Area of Science:

• Materials Science
• Chemical Engineering
• Artificial Intelligence

Background:

• Molding is crucial for aerospace and automotive thermoset composites, impacting cost and part reduction.
• Evaluating resin solidification requires the degree-of-cure curve, but simulations need accurate initial conditions and face computational demands.
• Coupled heat transfer and cure kinetics simulations are time-intensive, highlighting the need for efficient predictive tools.

Purpose of the Study:

• To develop a data-driven artificial intelligence approach for predicting the degree-of-cure curve in molded composite materials.
• To compare the predictive accuracy of different machine learning models for cure behavior.

Main Methods:

• Finite element simulations generated temperature-time and degree-of-cure-time data.
• Machine learning models, including Support Vector Regression (SVR), Back Propagation (BP) neural network, and a Genetic Algorithm-optimized BP (GA-BP) network, were trained on simulation data.
• Model performance was validated using evaluation indices.

Main Results:

• The simulated degree-of-cure curve for a specific temperature-time profile was verified against published data.
• The GA-BP neural network model achieved the highest accuracy in predicting the degree-of-cure curve compared to SVR and standard BP.
• Validation indices confirmed the superior performance of the GA-BP model.

Conclusions:

• A data-driven approach using machine learning, particularly the GA-BP network, is effective for accurately predicting the degree-of-cure in composite materials.
• This AI-powered prediction can significantly improve the efficiency of simulating composite curing processes.
• The developed model offers a promising solution for optimizing manufacturing processes in the aerospace and automotive industries.