263134 research outputs found
Sort by
Ontwikkeling van Procesbewaking en Real-time Defectdetectie voor Wire-Arc Additive Manufacturing
No full textWire Arc Additive Manufacturing (WAAM) is a promising technology for fabricating large and complex metal components with reduced material waste and improved efficiency. To fully realize its industrial potential, ensuring consistent build quality through robust monitoring is essential. However, real-time process monitoring remains a challenge due to the complexity of the underlying welding physics and the occurrence of defects such as porosity, cracks, and geometrical deviations.
This thesis advances WAAM monitoring through four key contributions. First, a multi-modal, high-resolution sensing system was developed, integrating acoustic, electrical, and optical modalities. The change of the data acquisition system from 500 kHz to 2 MHz enabled high-fidelity capture of transient events such as spatter and bubble bursts, with synchronized high-speed video providing robust physical validation. Second, a systematic comparison of acquisition rates revealed that while 2 MHz sampling allowed the use of high-frequency acoustic emission sensors, these did not improve defect classification. Instead, microphones operating in the 500 Hz-10 kHz range consistently provided the most informative signals, demonstrating that effective defect detection does not require extreme sampling rates. Third, an interpretable modeling framework—Sparse Wavelet Networks (SWaN) was introduced, combining wavelet scattering transforms with sparse principal component analysis to retain classification accuracy while highlighting physically meaningful features linked to melt pool dynamics and process stability. An optimized lightweight version (SWaN-Lite) further reduced computational overhead, making real-time monitoring feasible. Finally, the monitoring system and models were validated in laboratory and industrial settings, including case studies in laser cladding, demonstrating both robustness and transferability.
Together, these contributions provide a sensor-integrated, interpretable, and real-time capable framework for WAAM defect detection, bridging the gap between high-performance data-driven modeling and the transparency required for industrial adoption.status: Publishe
Ultrasone Geleide Golf Structurele Gezondheidsmonitoring: Een Probabilistisch Deep Learning Perspectief
No full textIntro: Structural Health Monitoring (SHM) is a vital technology in engineering design and operational management focused on ensuring structural efficiency and safety. It plays an essential role in extending structural service life, minimizing maintenance costs, and preventing catastrophic failures. A central objective of SHM is the early detection of structural anomalies before they lead to critical damage, thereby enabling preventive inspection strategies. Ultrasonic Guided Waves (UWGs), which are boundary-guided, exhibit low attenuation over long distances and are highly sensitive to damage and defects—even minor damage can lead to observable changes in wavefield features. As a result, UGW monitoring has become one of the most widely adopted non-destructive techniques in SHM. Deep learning has recently emerged as a popular research method across various SHM applications and is expected to play an increasingly pivotal role in the future. By employing deep learning models for feature extraction, anomaly detection, and predictive analysis of UGW data from damaged structures, SHM systems can achieve higher levels of automation, improved real-time monitoring, and enhanced capacity for managing complex data.
Why: Despite recent advancements, SHM technologies that integrate traditional deep learning with UGWs continue to face several significant challenges: (i) effectively extracting damage-sensitive features from UGWs; (ii) traditional deep learning models are unable to provide statistical grounded confidence estimates for their predictions; (iii) the difficulty of acquiring high-quality labeled data in many practical settings; (iv) limited generalizability of models trained in one domain or scenario when applied to others. To enable the practical deployment of UGW-based SHM systems, it is essential to address four interconnected challenges: feature reliability, uncertainty, label scarcity, and domain shift.
How: To address the core challenges, this work proposes a systematic technical framework: feature engineering→probabilistic deep learning→probabilistic unsupervised transfer learning. Specifically,
• Feature engineering:
(a) Data level: A various UGW database is developed via numerical simulations and experimental measurements, encompassing four scenarios: delamination in composite beams, microcracks of 3D-printed auxetic honeycomb structures, surface defects in aluminum-stiffened panels, and corrosion in underwater steel pipes.
(b) Feature level: A novel five-dimensional systematic evaluation criterion is introduced to quantitatively assess 15 damage-sensitive features across time, frequency, and time-frequency domains.
• Probabilistic deep learning: Enhancements to the VGG-13 backbone with probabilistic layers to form three probabilistic deep learning models—Flipout probabilistic convolutional neural network (FPCNN), deep ensemble PCNN, and Bayesian PCNN—each integrating uncertainty-aware components and leveraging techniques like Flipout variational inference, Bayesian layers, and deep ensembles. These models provide not only diagnostic outputs but also 95% confidence intervals capturing both aleatoric and epistemic uncertainties.
• Probabilistic unsupervised transfer learning: Two unsupervised transfer learning frameworks are proposed. The first, Multi-scale Adaptive Attention Transformer Domain Adaptation Network (MAAT-DAN), combines a multi-scale adaptive attention Transformer with adversarial feature alignment and maximum mean discrepancy regularization to achieve domain adaptation without target domain labels. The second, Deep-Ensemble Unsupervised Transfer-Learning (DE-UTL), integrates deep ensembles with Wasserstein distance to further improve the same label-free adaptation while also quantifying both types of uncertainty.
Results: The constructed multi-dimensional feature evaluation framework provides a reference for processes of feature selection and evaluation. Furthermore, the proposed probabilistic deep learning framework demonstrates strong generalization capabilities by maintaining low prediction errors across diverse conditions, including different materials (composites, auxetic structures, aluminum alloys, steel pipes), environments (from air to water), and intersensor/intertemporal (from 2012 to 2013).
(1) In feature evaluation, this work proposes five criteria—diagnostic capability, environmental stability, computational complexity, sensor configuration, and structural applicability—and applies them to classify 15 common features into four performance levels.
(2) In the composite delamination dataset, the Flipout-based network achieves an optimal balance between training efficiency, uncertainty quantification effectiveness, and accuracy in localizing damage.
(3) In 3D-printed auxetic structures, FPCNN consistently provides compression deformation predictions within the 95% confidence interval of total uncertainty. It is also capable of early identification of critical damage at the point of a sharp drop in envelope energy.
(4) In metal structures, MAAT-DAN significantly outperforms conventional models in unsupervised cross-material regression and cross-temporal sensor classification tasks, showcasing its robustness in feature extraction, domain alignment, and stable generalization across domains.
(5) In underwater structures, DE-UTL demonstrates high accuracy in unsupervised corrosion thickness diagnosis and uncertainty quantification under geometry-environmental transfer from a rectangular cross-section steel pipe in air to a square cross-section pipe in water.
Overall, the experimental results confirm that the proposed guided wave-based probabilistic deep learning SHM framework achieves high accuracy and reliability in scenarios involving different materials, harsh environments, and task switching. It significantly reduces the need for labeled data and potentially improves engineering deployability. This work establishes a comprehensive technical roadmap for UGW SHM—from feature engineering to probabilistic inference, and unsupervised transfer learning—laying a solid theoretical and practical foundation for future intelligent, reliable, cross-scenario, and cost-effective SHM systems.status: Publishe
A study on water resistance of polyurethane binder for pavements through formula optimization and coupling environmental factors
No full textWith increasing requirement for engineering performance and functionalities in the pavement, polyurethane (PU) binder is proposed to replace asphalt binder due to its superior mechanical properties and curtailed formula. However, water susceptibility of PU binder is a critical issue to limit its wide application. This study aims to optimize chemical formula of PU binder with superior water resistance and evaluate its water susceptibility under the coupling effects of multiple environmental factors. It has revealed that the type of isocyanate has little effect on water contact angle and water absorption of PU. As molecular weight (Mn) of polytetramethylene ether glycol increases from 650 to 2000 and the ratio of hard segment and soft segment (R value) rises from 1.8 to 2.2, PU binder presents larger water contact angle and lower water absorption while decreased mechanical properties. With Mn of 1000 and R value of 2.2, PU binder features superior water resistance, which exhibits water contact angle of 95.7°, water absorption of 0.06 %, tensile strength of 18 MPa and breaking elongation of 266 %. Under the coupling effect of temperature and water immersion, tensile strength of PU samples maintains over 73 % and 54 % as temperature is above 55 °C and 85 °C, respectively, compared to original PU samples. Under the coupling effect of temperature and relative humidity, tensile strength of PU samples is degraded less than 15 % at temperature of 55 °C and relative humidity of 90 %. Under the coupling effect of water immersion with various pH values, tensile strength of PU samples remains over 75 % and 80 % at pH value of 5 and 9, respectively compared to original PU samples. Moreover, under coupling environmental factors, PU binder with optimized formula exhibits negligible change in structural morphology, glass transition temperature and chemical groups. The research outcomes facilitate design and application of PU binder with desirable water resistance in the pavement.sponsorship: This research was supported by the National Natural Science Foundation of China (Grant Nos. 6521009749, 51922030) , SEU Innovation Capability Enhancement Plan for Doctoral Students (CXJH_SEU 25190) . Thanks to the anonymous reviewers for constructive comments and suggestions which could help us improve the manuscript. (National Natural Science Foundation of China|6521009749, National Natural Science Foundation of China|51922030, SEU Innovation Capability Enhancement Plan for Doctoral Students|CXJH_SEU 25190)status: Publishe
Divided we stand: a fiscal bargaining model for heterogeneous countries
No full textstatus: Published onlin
The impact of work-based versus school-based learning on cognitive and non-cognitive outcomes in vocational secondary education
No full textWe investigate how work-based versus school-based learning impacts students’ cognitive (numeracy, literacy)
and non-cognitive outcomes (student motivation, engagement, academic self-concept, well-being) in vocational secondary education. To this end, we exploit longitudinal test score and survey data on a cohort of Belgian pupils. We rely on school-field-of-study fixed effects, (non-)cognitive test scores in the preceding grades, and detailed background characteristics to remove the bias resulting from self-selection into educational programmes. Within the technical track aimed at preparing students both for the labour market and further education, we find the effects of substituting school-based for work-based learning on both cognitive and non-cognitive outcomes to be negative. Within the purely vocational track, by contrast, effects seem to be more mixed and to depend on the outcome and amount of workplace learning. These results are consistent with
cognitive and non-cognitive outcomes being crucial mechanisms underlying the effects of work-based pro-
grammes on educational and labour market career outcomes.sponsorship: This work was supported by the Flemish Authority within the frame of the policy research centre on educational research (Steunpunt SONO) . Kristof De Witte acknowledges financial support from the European Commission through its Horizon Europe project BRIDGE (grant 101177154) . Views and opinions expressed are however those of the author (s) only and do not necessarily reflect those of the European Union or the granting authority. Neither the European Union nor the granting authority can be held responsible for them. (European Commission through its Horizon Europe project BRIDGE|101177154, Flemish Authority within the frame of the policy research centre on educational research (Steunpunt SONO))status: Published onlin
Numerical study of the impact of global mechanisms in LES of propane pool fire using the EDC-finite-rate chemistry approach
No full textsponsorship: This paper is written as the extension of the earlier publication [49] in proceedings of the 4th European Symposium on Fire Safety Science held from 9 to 11 October 2024 in Barcelona, Spain. This research is funded by The Research Foundation-Flanders (FWO-Vlaanderen) via the senior research project G023221N, G034725N, and PhD fellowship 1104125N. (Research Foundation-Flanders (FWO-Vlaanderen)|G023221N, Research Foundation-Flanders (FWO-Vlaanderen)|G034725N, Research Foundation-Flanders (FWO-Vlaanderen)|1104125N)status: Accepte
Betere zorg voor chronische rhinosinusitis patiënten door het onderzoeken van nieuwe therapieën die de mucosale epitheliale barrière beïnvloeden
No full textThe current treatment options for patients suffering from chronic rhinosinusitis (CRS) are limited, hence leading to suboptimal disease control. Better understandingk of the pathophysiology and implementing of novel preventive treatment strategies is warranted. The current PhD project aims at improving the current understanding of efficacy of novel treatment optons with biologicals for CRSwNP, and to evaluate efficacy of antiviral strategies in a preventive context. Studying the effects of different biologicals on nasal epithelial barrier dysfunction, transient receptor potental expression in the (sino)nasal mucosa and systemic biomarkers, comparing the outcomes of different biologicals on clinical and immunologic features of CRS, and comparing outcomes of biological treatment with sinonasal surgery will all pave the future of CRS care. In additon, novel approaches preventing the virus-induced nasal epithelial barrier dysfunction will be studied in the context of CRS exacerbation and prevention Outcomes of this PhD transaltional research programme will have an impact on the clinical practice of patients suffering from CRS.status: Publishe
Results of the Ninth Scientific Workshop of the European Crohn's and Colitis Organisation (ECCO): Artificial intelligence in IBD surgery: opportunities and limitations.
No full textIn this narrative review we present the current status of developments in artificial intelligence in the filed of IBD surgery. We lay down the foundations for how IBD surgery may utilise the potential opportunities in utilizing the rapid advances in AI technology as it used in other surgical disciplines. The main areas of potential utility are in the areas of surgical training, risk prediction in the pre, intra and post operative period in IBD patients undergoing surgery and in IBD surgical research. We need to be mindful of the potential challenges in implementation and acceptability of these technological advances and put in mitigating measures to ensure transparency and equitable access. Global collaboration will be the cornerstone for such ventures.status: Published onlin
Nervus peroneus entrapment: nieuwe inzichten in diagnose en behandeling.
No full textPeroneal nerve entrapment is the most common neuropathy in the lower limbs and a common cause of foot drop. Patients with foot drop have gait difficulties and an increased risk at falling. Daily practice is very variable, despite the frequency of the pathology. Some centers tend to operate the patients in an early stage whereas other centers never operate and treat the patients in a conservative manner. This polarisation in daily practice is a reflection of a lack of evidence in the current literature. The literature consists mostly of small, retrospective case series. Few prospective trials / patients series are available and randomized controlled trials have not been conducted so far. There is little to no epidemiological data available. The purpose of this doctoral proposal is to conduct a multicenter, randomized controlled trial to determine the superior treatment of peroneal nerve entrapment. These results can be translated in guidelines for daily practice. Epidemiological and electrophysiological data will be collected as well. 24 Belgian centers and 2 Dutch centers will participate in this trial.status: Publishe
Magnetostatic and Magnetodynamic Modeling with Unsupervised Physics-Informed Neural Networks
No full textsponsorship: This work was supported by Flemish Government through InduFlexControl-2: Control Algorithms for Flexibility in Power-to-X and Industrial Processes-2 under Project HBC.2021.0579. (Flemish Government through InduFlexControl-2: Control Algorithms for Flexibility in Power-to-X and Industrial Processes-2|HBC.2021.0579)status: Published onlin