Physics Informed Neural Network (PINN) Utilization in Discovery of Diffusion–Electrical Cross–property Relation in Polymer Composites

This research work investigates the relationship between moisture absorption and changes in electrical properties in polymer and polymer matrix composites. The dielectric properties of the matrix are altered due to induced polarization mechanisms by water molecules. Fick's law of diffusion can be used to model moisture absorption, and Maxwell's equations of electromagnetism can be used to model the electrical response of the material. However, computationally coupling diffusion and Maxwell's equations in hygrothermal loading of polymers remain unexplored. A novel approach using physics-informed neural networks (PINN) is proposed to address this. The proposed methodology explores the coupling between diffusion and evolving electrical properties through the development of a physics informed deep neural network framework. The proposed method has the potential to estimate moisture diffusion based on electrical properties, and vice-versa, providing insights for composite structure design and durability assessment in a non-conservatism manner.