Integrating microscale and macroscale simulations has long been a computational challenge in materials science. To address this, the researchers developed a machine learning model that efficiently predicts the behavior of materials used in wearable electronics, his AGAT with a particular focus on CNT/PDMS composites. .
The AGAT model was established based on the architecture of an artificial neural network (ANN) incorporating a graph attention network. The embedded ANN layer is used to predict the Young's modulus from the CNT length and radius, serving as a bridge between molecular and mesoscale models.
Utilizing extensive multiscale simulations and data from existing literature, a model is trained to evaluate the material sensing properties of CNT/PDMS composites with high accuracy.
The AGAT model significantly reduces computational overhead for material properties essential for flexible electronic devices. By bridging the gap between detailed molecular simulations and practical macroscale applications, this model enables designers to explore new materials and optimize them for electronic interfaces with high efficiency.
Research results will be published in a magazine national science open. The research was led by Professor Xiaonan Wang and Dr. Lingjie Yu.
For more information:
Lingjie Yu et al., Machine learning-enabled approach for bridging multiscale simulation of CNT/PDMS composites, national science open (2024). DOI: 10.1360/nso/20230055
