MPCA-based Domain Adaptation for Transfer Learning in Ultrasonic Guided Waves

2025-08-01

AI Summary

Ultrasound guided wave (UGW)-based structural health monitoring (SHM) suffers from data scarcity and poor generalization across diverse materials and sensor configurations. Method: This paper proposes a transfer learning framework leveraging multilinear principal component analysis (MPCA) to jointly perform tensor dimensionality reduction and extract shared latent features from source and target domains—without requiring predefined dimensions or domain-specific priors. A convolutional neural network (CNN) is employed for damage localization regression, followed by lightweight fine-tuning to adapt to new platforms, substantially reducing reliance on labeled target-domain data. Results: Evaluated across 12 composite material–sensor array combinations, the method achieves significant inter-domain feature alignment, reducing localization error by 37.6% on average compared to conventional transfer approaches. It markedly enhances cross-domain robustness and engineering deployability of UGW-SHM models.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Computer Vision: Multi-modal Vision

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Web Mining and Content Analysis: Mining multimedia, multimodal, multilingual, cross-lingual Web data

Abstract

Ultrasonic Guided Waves (UGWs) represent a promising diagnostic tool for Structural Health Monitoring (SHM) in thin-walled structures, and their integration with machine learning (ML) algorithms is increasingly being adopted to enable real-time monitoring capabilities. However, the large-scale deployment of UGW-based ML methods is constrained by data scarcity and limited generalisation across different materials and sensor configurations. To address these limitations, this work proposes a novel transfer learning (TL) framework based on Multilinear Principal Component Analysis (MPCA). First, a Convolutional Neural Network (CNN) for regression is trained to perform damage localisation for a plated structure. Then, MPCA and fine-tuning are combined to have the CNN work for a different plate. By jointly applying MPCA to the source and target domains, the method extracts shared latent features, enabling effective domain adaptation without requiring prior assumptions about dimensionality. Following MPCA, fine-tuning enables adapting the pre-trained CNN to a new domain without the need for a large training dataset. The proposed MPCA-based TL method was tested against 12 case studies involving different composite materials and sensor arrays. Statistical metrics were used to assess domains alignment both before and after MPCA, and the results demonstrate a substantial reduction in localisation error compared to standard TL techniques. Hence, the proposed approach emerges as a robust, data-efficient, and statistically based TL framework for UGW-based SHM.

Problem

• Addresses data scarcity in UGW-based ML methods
• Improves generalization across materials and sensor configurations
• Enables domain adaptation without large training datasets

Innovation

• MPCA-based domain adaptation for transfer learning
• Convolutional Neural Network with fine-tuning
• Shared latent feature extraction across domains