Newsfeeds
Measurement Science and Technology - latest papers
Latest articles for Measurement Science and Technology
-
DGCDN: robust acoustic fault diagnosis via domain-generalized causal disentanglement
Reliable industrial fault diagnosis depends on the robust interpretation of sensor measurements, which are often degraded by unseen operating conditions. Acoustic sensing, a flexible and non-contact modality, is highly desirable for this purpose; however, its practical application is severely hindered by low signal fidelity and pervasive background noise. These measurement challenges introduce significant domain shifts that current domain generalization (DG) methods struggle to overcome, often failing to disentangle faint physical fault signatures from nuisance factors inherent in the measurement environment. To address these measurement-centric limitations, we propose the domain-generalized causal disentanglement network (DGCDN). Grounded in a structural causal model of the signal generation and measurement process, DGCDN learns domain-invariant health representations by causally isolating them from domain-specific measurement artifacts. This disentanglement is enforced through a synergistic multi-objective function that preserves measurement fidelity via reconstruction, ensures feature independence through an orthogonality constraint, and promotes domain invariance using contrastive learning. The training process is further stabilized by an adaptive domain weighting mechanism and entropy regularization, which respectively prioritize challenging source domains and mitigate predictive uncertainty. Validation on datasets exhibiting substantial shifts in measurement conditions confirms the model’s efficacy, as it consistently achieves state-of-the-art performance across a suite of evaluation metrics, including accuracy, precision, recall, F1-score, and AUC. These results highlight DGCDN’s exceptional robustness to domain shifts and data scarcity, underscoring its potential for reliable deployment in high-stakes industrial measurement systems.
-
Single-view iterative measurement of rotary axis radial error motion utilizing line-structured light
Currently, optical measurement technology has distinct advantages in evaluating the radial error of the axis of rotation. However, a significant challenge persists: how to rapidly characterize this error from the spherical point cloud captured at a single viewing angle. As a symmetric geometric object, the reference sphere necessitates precise calibration of the structured light plane to its center to extract radial runout point clouds, and the calibration is often intricate and time-consuming. Moreover, installation eccentricity and random optical noise increase the complexity of accurately extracting radial runout data. To overcome these issues, a novel single-view iterative measurement framework (SVIMF) is proposed for the first time to enable rapid characterization of radial errors in eccentric shafts. The SVIMF comprises four primary modules: calibration, parameter adjustment, measurement, and evaluation. A three-step centering model based on reference sphere features is developed to determine the optimal measurement position within the structured light field of view. Furthermore, a radial runout point cloud reconstruction methodology is proposed, and a detection framework correlating the radial dimensional variation of the rotation axis with the radial runout point cloud established. Finally, the Fourier transform is employed for harmonic decomposition and synchronous error extraction, thereby enabling the quantitative characterization of the radial error of the rotation axis. Experimental results validate the feasibility and substantial application potential of the proposed SVIMF.
-
A dynamic spatial-temporal graph transformer with multi-frequency attention for remaining useful life prediction
The remaining useful life (RUL) prediction is essential for cost-effective production and reliable predictive maintenance in intelligent manufacturing. Existing deep learning-based approaches often struggle to capture complex degradation patterns across temporal, spatial, and frequency domains. To address this limitation, a dynamic spatial-temporal graph transformer with multi-frequency attention (DSTGT-MFA) is proposed in this paper for RUL prediction. The proposed DSTGT-MFA model consists of three key components: a multi-scale gated convolutional neural network for extracting hierarchical local features, a graph convolution transformer for modeling long-term spatial-temporal dependencies with dynamic and static adjacency matrices, and a multi-frequency spatial-temporal attention mechanism to enhance temporal and spatial attention in the frequency domain. This integrated architecture enables the model to comprehensively capture degradation trends and fuse multi-domain features. Extensive experiments conducted on the commercial modular aero-propulsion system simulation (CMAPSS) and new CMAPSS datasets demonstrate that the DSTGT-MFA model achieves superior prediction accuracy compared to twelve baseline methods.
-
Adaptive transfer learning network optimized by an improved snow ablation algorithm for cross-condition gearbox fault diagnosis
When gearboxes operate under variable conditions, their vibration signals often exhibit significant distribution shifts, which compromise the accuracy and reliability of fault diagnosis. Existing methods face three major limitations: (1) conventional domain adaptation approaches are prone to domain and class confusion; (2) traditional feature extractors fail to comprehensively capture fault characteristics; and (3) hyperparameter optimization algorithms in high-dimensional spaces are easily trapped in local optima. To address these challenges, this work proposes an adaptive transfer learning network (ATLN) optimized by an improved snow ablation optimizer (ISAO). The ATLN employs a convolutional neural network–gated recurrent unit framework to efficiently extract spatiotemporal features from vibration data, while an enhanced joint distribution adaptation strategy with a new hybrid distance metric – maximum earth discrepancy (MED) reduces distribution misalignment across domains. Furthermore, the Enhanced Transfer Softmax (ET-Softmax) layer incorporates Center Loss and KL divergence to enhance intra-class compactness and inter-class separability. Experimental results on three open gearbox datasets show that ATLN (ISAO) achieves an average accuracy of 97.31% under multiple operating conditions and consistently maintains over 90% accuracy under compound conditions. These results demonstrate the robustness and strong cross-domain generalization of the proposed framework, enabling accurate and reliable gearbox fault diagnosis from vibration data under complex operating conditions.
-
Fast neutron spectroscopy with 4H-SiC solid-state detectors up to 500 °C for nuclear fusion applications
Silicon carbide (SiC)-based detectors offer exceptional radiation hardness and thermal stability, making them suitable for neutron spectroscopy in fusion reactor environments, which are characterized by high temperatures and intense neutron fluxes. In this study we demonstrate a 250 µm-thick 4 H-SiC p–n junction detector that maintains stable deuterium–tritium neutron detection performance across the full temperature range from 25 °C to 500 °C, thereby overcoming the limitations commonly encountered with diamond-based detectors. These results highlight the potential of thick SiC detectors for monitoring neutron flux and performing neutron spectroscopy in harsh environments, such as the breeding blanket of fusion reactors.