Neural Operator-Based Damage Sensing Digital Twin

• Developed with NeuralOperator, a comprehensive PyTorch library for learning neural operators, containing the official implementation of Fourier Neural Operators and other neural operator architectures.
• Coupling a machine learning-enabled multiscale multiphysics model, viz. the Parametrically Upscaled Coupled Constitutive Damage Model (PUCCDM) with Fourier neural operator (FNO) learning, this work develops a Virtual Damage Sensing Digital Twin (VDS-DT) platform for piezoelectric composites. This platform interfaces with actual surface sensor measurements for efficient prediction of full-field temporally evolving deformation, electrical, and damage fields in the entire structure.

Gallery

Hierarchical structure of a Fourier neural operator (FNO)-derived Virtual Damage Sensing Digital Twin (VDS-DT)for piezoelectric composites (Liu and Ghosh 2026)

Input for operator learning (Liu and Ghosh 2026)

Output for operator learning (Liu and Ghosh 2026)

References

Liu, Yangyuanchen, and Somnath Ghosh. 2026. “Neural Operator-Based Damage Sensing Digital Twin for Piezocomposites with Parametrically Upscaled Coupled Constitutive Damage Model.” Computer Methods in Applied Mechanics and Engineering 461: 119169. https://doi.org/https://doi.org/10.1016/j.cma.2026.119169.