論文

This study proposes a machine learning-based framework for predicting three-dimensional internal defects in a complex-shaped carbon fiber reinforced plastic (CFRP) specimen, using experimentally measured thermoelastic and computational FEA stress distributions. The proposed method employs a graph neural network (GNN), which directly incorporates the FEA mesh. First, high-frequency noise in the thermoelastic two-dimensional stress distribution is removed via Fourier analysis. Then the denoised field is mapped, using partial least squares and Gaussian process regression, into the principal component analysis (PCA) latent space of two-dimensional FEA stress distributions. Next, a multi-layer perceptron projects the latent variables of the two-dimensional stress fields into the latent space of the three-dimensional FEA stress distributions. The resulting latent variables are then reconstructed via PCA to obtain the three-dimensional stress distribution. Finally, the reconstructed three-dimensional stress distribution is input to a GNN trained on FEA datasets to predict three-dimensional internal defects. Conventional nondestructive inspection techniques for CFRP structures, such as X-ray imaging and ultrasonic testing, face safety, cost, and geometric constraints. In contrast, infrared stress measurement offers safer, simpler, and real-time inspection within the field of view. The proposed method estimates internal defect positions from surface stress information and provides a basis for reducing expert-dependent interpretation in future workflows. The present study demonstrates the feasibility of the proposed framework using one complex-shaped CFRP specimen with curved surfaces, such as a prosthetic-leg geometry, under the prescribed defect conditions.

Self-consistent field (SCF) analysis is an indispensable tool for predicting the microphase separation structures of polymer alloys. However, the computation of the phase-separated structures in the equilibrium state is computationally intensive, leading to high costs. To address this challenge, we propose a novel deep learning approach that leverages a generative adversarial network (GAN), a powerful deep generative model, to accelerate SCF analysis. Specifically, we trained the GAN using comprehensive data obtained from SCF analysis, enabling us to generate various images of feasible structures that are similar to the SCF analysis results. Our results demonstrate that the latent variables in the GAN are linked to the physical parameters and features of the phase-separation structures.

In this study, we propose a novel defect localization method that integrates the graph neural network (GNN) with the finite element method (FEM) to estimate the three-dimensional location of defects in perforated carbon-fiber-reinforced plastic (CFRP) interstage structures. Specifically, the model uses distributions of the sum of principal stresses on the surface (DSPSS) to predict the three-dimensional location of defects. FEM is employed to simulate tensile loading conditions and generate stress distribution data using Teflon sheets to represent predefined delaminations. These distributions serve as inputs to the graph attention network (GAT), which classifies defect positions into 19 categories. The proposed method achieved a macro-averaged F1-score of 61% and accurately predicted both the insertion layers and planar positions of defects.