A Multi-Fidelity Convolutional Autoencoder-Transfer Learning Framework for Guided-Wave-Based Damage Diagnosis Using Large Simulated and Limited Experimental Datasets

2026-06-25

AI Summary

This study addresses the challenges of scarce experimental data and high computational cost of high-fidelity simulations in guided-wave-based structural health monitoring by proposing a multi-fidelity transfer learning framework. The approach leverages a lightweight one-dimensional time-domain spectral element method to generate synthetic data for pretraining a convolutional autoencoder (CAE), whose deep features are subsequently integrated with a feedforward neural network. Fine-tuning with only a small amount of experimental data enables highly accurate damage localization and size estimation. This work represents the first application of CAEs to guided-wave damage diagnosis, demonstrating significantly enhanced generalization under extremely limited experimental data. The method achieves R² scores exceeding 0.93 for damage location and 0.99 for size prediction, outperforming conventional CNNs and confirming its accuracy, efficiency, and practical engineering utility.

Abstract

Guided wave-based structural health monitoring (GWSHM) with onboard transducers offers significant potential for the early diagnosis of damage in engineering structures. However, the practical deployment of deep learning models is often hindered by the limited availability of labelled experimental data and the high computational cost of generating large-scale high-fidelity simulation datasets. This study presents a multifidelity transfer learning framework that integrates lightweight physics-based simulations, convolutional autoencoder (CAE)-based deep feature learning, a feed-forward neural network, and limited experimental measurements for accurate damage localisation and sizing in plate-like structures instrumented with piezoelectric transducers. A computationally efficient one-dimensional time-domain spectral element model is employed to generate a large synthetic dataset for pretraining, while transfer learning adapts the model to experimental domains using only a small amount of labelled data. The CAE-based transfer learning framework significantly outperforms its CNN-based counterpart in damage localisation accuracy. The model achieves excellent predictive performance with $R^2$ scores exceeding 0.93 for damage localisation and 0.99 for damage sizing. Its generalisation capability is demonstrated on previously unseen data, showing high prediction accuracy for damage scenarios not represented during pretraining or fine-tuning. The results establish the proposed framework as an accurate, computationally efficient, and practically viable solution for real-world GWSHM applications.

Problem

• guided-wave-based structural health monitoring
• limited experimental data
• high-fidelity simulation
• damage diagnosis
• data scarcity

Innovation

• multi-fidelity
• convolutional autoencoder
• transfer learning
• guided-wave SHM
• physics-informed simulation