On the Condition Monitoring of Bolted Joints through Acoustic Emission and Deep Transfer Learning: Generalization, Ordinal Loss and Super-Convergence

TL;DR

This work tackles condition monitoring of bolted joints using acoustic-emission data and deep transfer learning. By converting AE signals to continuous wavelet transform scalograms and leveraging pretrained CNNs, it investigates generalization across measurement campaigns, ordinal loss formulations, and the rapid convergence offered by the 1cycle learning-rate schedule. The study finds that ordinal losses, particularly POM1b, and a non-freezing training regime with EfficientNetB5 or ResNet18 achieve strong cross-campaign performance, while the mu80 sensor alone suffices and denoising provides no benefit. Overall, the approach demonstrates robust, generalizable SHM capability for bolted joints and informs sensor choice, loss design, and training efficiency for real-world deployment.

Abstract

This paper investigates the use of deep transfer learning based on convolutional neural networks (CNNs) to monitor the condition of bolted joints using acoustic emissions. Bolted structures are critical components in many mechanical systems, and the ability to monitor their condition status is crucial for effective structural health monitoring. We evaluated the performance of our methodology using the ORION-AE benchmark, a structure composed of two thin beams connected by three bolts, where highly noisy acoustic emission measurements were taken to detect changes in the applied tightening torque of the bolts. The data used from this structure is derived from the transformation of acoustic emission data streams into images using continuous wavelet transform, and leveraging pretrained CNNs for feature extraction and denoising. Our experiments compared single-sensor versus multiple-sensor fusion for estimating the tightening level (loosening) of bolts and evaluated the use of raw versus prefiltered data on the performance. We particularly focused on the generalization capabilities of CNN-based transfer learning across different measurement campaigns and we studied ordinal loss functions to penalize incorrect predictions less severely when close to the ground truth, thereby encouraging misclassification errors to be in adjacent classes. Network configurations as well as learning rate schedulers are also investigated, and super-convergence is obtained, i.e., high classification accuracy is achieved in a few number of iterations with different networks. Furthermore, results demonstrate the generalization capabilities of CNN-based transfer learning for monitoring bolted structures by acoustic emission with varying amounts of prior information required during training.

Figures (14)

• Figure 1: ORION-AE benchmark and setup configuration ORIONdata.
• Figure 2: Diagram depicting contact conditions of the Orion beam at interface, and a microscale contact area illustration which highlights the contact state at each campaign.
• Figure 3: Workflow containing three different modules: (1) Signal Processing, (2) Data Preparation and (3) Tightening Level Identification.
• Figure 4: Schematic representation of the studies carried out during the damage detection stage, considering different network architectures, batch sizes, loss functions, fine-tuned or frozen layers, learning rate scheduler and optimizer. In this figure, SGD stands for Stochastic Gradient Descent.
• Figure 5: Effect of apodization with a Hanning window on the CWT.