139585 research outputs found
Sort by
Data-driven torsional vibration-based fault diagnosis of large internal combustion engines without real fault data
No full textCylinder-specific condition monitoring is critical for ensuring reliability and safety in large combustion engines. However, conventional approaches typically require extensive instrumentation, resulting in high implementation costs and complexity. This paper presents a data-driven and sensor-efficient condition monitoring methodology for large industrial engines that achieves accurate fault detection using only a single flywheel encoder measurement. Despite the scarcity of real-world fault data, the framework leverages simulation-based training enhanced by domain randomization, feature alignment, and semi-supervised learning techniques to bridge the simulation-to-reality gap. A modified Deep Convolutional Neural Network with Wide First-layer Kernels (WDCNN) is employed for robust fault classification. The framework is validated on a 20-cylinder gas engine. Compared to conventional lateral vibration- or pressure-based monitoring systems that rely on distributed multi-sensor frameworks, this approach achieves comparable accuracy with drastically reduced sensor requirements, reducing the required amount from up to 40 to 1. A lower number of required sensors maintain higher reliability levels, since higher sensor counts expose the system to a higher rate of sensor failure. Experimental results show 100% fault detection accuracy and 95.7% classification accuracy on a dataset consisting of limited real measured data, highlighting the framework’s potential for practical deployment in real-world industrial settings with minimal sensor setups. Validation was conducted using a single real-world fault condition, underscoring the need for future validation on broader fault generalization. Nevertheless, this work demonstrates the feasibility of high-precision engine fault monitoring with dramatically reduced sensor requirements, enabling cost-effective diagnostics in data-scarce industrial environments.Peer reviewe
Learning cultural profiles on collaborative learning: the case of Finnish multi-platform youth content
No full textBy experimenting with new kinds of working arrangements, public broadcasters attempt to ensure their future audiences while facing heavy competition. This article considers the role of collaborative learning in audience retention by comparing the methods and views of audiovisual workers from differing media cultures and diverse generational age groups. Drawing on the theoretical-methodological frameworks of organisational learning and developmental work research, the study identifies four different learning cultural profiles. The ethnographic data is collected from a multi-platform youth campaign of the Finnish Broadcasting Company Yle. It included employees’ own reflections on their learning. According to the findings, Yle’s employees recognised most of the characteristics of innovative learning culture. Yle’s audiovisual specialists from the studio production team were keen to serve the audience, although they were left aside from the core social media team. Media employees with fixed-term contracts were in the most vulnerable position in collaborative learning. Especially the young beginners from the commercial production company had no means of questioning their subordinate status. The article suggests that in moving towards a more co-creational working culture, becoming acquainted with the audience should be a shared strategic objective for the whole collaborative media production team.Peer reviewe
Bionic fusion perspective : Audiovisual-motivated integration network for solar irradiance prediction
No full textPublisher Copyright: © 2024 Elsevier LtdAccurate and reliable prediction of solar irradiance (SI) is an important requirement to develop solar energy while a challenging task due to stochastic and nonlinear data characteristics. Additionally, most deep networks show powerful prediction capabilities but lack the supports from biological science, reflecting that bionically-inspired networks in SI analysis are still not enough explored. To this end, this paper proposes an audiovisual-motivated Transformer-CNN integration network, called ATI-net, for predicting SI. The audiovisual cognition gives a superior design framework for ATI-net with signal capture, signal analysis, and prediction blocks. In the first block, through mimicking the function of both eye and ear in external signal conversion, multi-scale features are extracted by incorporating multi-branch convolutions with varying kernels, where the Mish function addresses the problem that traditional ReLU function stops learning when the input is negative. In the second block, through mimicking the function of left and right hemispheres in neuronal signal analysis, two structures triggered by Transformers and convolutions are designed to remember temporal evolutionary rules, where residual connections are beneficial to mine deep information and avoid forgetting. In the third block, through mimicking the function of a higher brain region in generating understanding, the above information is integrated to make the SI prediction. Besides, the nonlinear dependencies and linear relationships are independently extracted and integrated into the ATI-net, which not only reduces information interference but is consistent with the “divide and conquer” idea. Experimental results show that the ATI-net outperforms 18 benchmarks, and average improvements of root mean squared error (RMSE) are 26.28% and 26.01% for two datasets, respectively. In summary, the ATI-net is one of the reliable alternatives to SI prediction.Peer reviewe
Influence of moisture on the sound absorption properties of wood-based pulp fibre foams
No full textPublisher Copyright: © 2024The use of wood fibres, and other biofibres in general, as raw materials to produce sound absorbers has gradually increased lately. The main reason for their increased use is their contribution to reducing CO2 from the atmosphere by binding CO2 into the building structure for decades. Understanding of the ultrastructure of wood fibres is essential as it has a strong influence on the fibre properties, and thus, on the final material properties. In this work, the effect of moisture on the sound absorption properties of wood-based pulp fibre foams has been studied. It is shown that increasing moisture content (> 9%) in pulp fibres leads to greater sound absorption at low-mid frequencies. Fibre swelling, increasing fibre flexibility, and increasing foam bulk density with increasing fibre moisture content are hypothesized as the causes for the increase in sound absorption. Hygroexpansion, mechanical properties as well as moisture absorption capability of different types of pulp fibres are studied and related to their sound absorption properties. It is concluded that, in addition to fibre diameter and bulk density of foams, the elastic properties of the pulp fibres are partially responsible for the improved sound absorption of the foams exposed to greater relative humidity conditions.Peer reviewe
Effect of In-Situ Catalyst on Co Extraction from Lithium-Ion Battery Scrap Via Selective Sulfation Roasting
No full textPublisher Copyright: © The Minerals, Metals & Materials Society 2025.Recycling of lithium-ion batteries has become essential to meet the demand for critical raw materials, especially lithium (Li) and cobalt (Co). Selective sulfation roasting followed by water leaching has been shown to be a promising route for recycling. However, it is very challenging to achieve high metal recovery efficiencies consistently with different batches of battery scrap because of their complex morphology and continuously changing chemistries. In this study, two different batches of lithium cobalt oxide (LCO)-rich black mass were treated by selective sulfation roasting and a water leaching process under similar conditions. The metal recovery efficiency of Li was observed to be consistently very high, whereas Co extraction efficiency was found to vary dramatically from one batch of black mass to another. It was demonstrated that this variation in metal extraction efficiency was mainly due to the considerably higher iron concentration in the first batch of black mass, where iron oxide acted as a catalytic agent and enhanced the formation kinetics of cobalt sulfate, resulting in higher Co recovery during water leaching. The catalytic effect of Fe2O3 was confirmed by demonstrating the enhancing effect on metal extraction efficiency by sulfation roasting with addition of Fe2O3 in LiCoO2 powder and low-iron black mass.Peer reviewe
Simulation of the buffer-IPyC interface cell from the TRISO nuclear fuel with LAMMPS
No full textPublisher Copyright: © 2025 The Author(s)The main advantage of the Tristructural Isotropic (TRISO) fuel is its four-layer coating, which prevents the release of fission products during reactor operation. Studies have shown that the behavior of the buffer-IPyC interface plays a crucial role in maintaining the integrity of the TRISO particle. In this study, we use molecular dynamics simulations to investigate the interface between these two structures and to provide insight into the atomic-level mechanisms responsible for irradiation-induced changes in the graphite-based structures. We identify the underlying mechanisms leading to swelling of the IPyC layer and reveal the effect of the interface on defect formation in the differing structures of the buffer and IPyC layers.Peer reviewe
B cell dysregulation during acute COVID-19 is transient
No full textPublisher Copyright: © 2025Background: COVID-19 is still a significant health concern worldwide. B cell responses to COVID-19 have been extensively studied in acute severe disease, but less so during extended follow-up or mild disease. Persisting immunological changes together with herpesvirus reactivations during acute COVID-19 have been suggested as contributing factors for post-acute sequelae of COVID-19 (PASC). Here, we evaluated the natural kinetics of B cell subpopulations together with serological markers of increased B cell activity during acute COVID-19 and long-term follow-up. We also measured human herpesvirus reactivations during acute COVID-19. Methods: We collected plasma and peripheral blood mononuclear cell samples from 120 SARS-CoV-2 positive patients (outpatients = 56, inpatients = 64) at up to five timepoints during acute disease and recovery (up to 460 days since symptom onset, dsso). We determined circulating B cell and Th cell subpopulations using flow cytometry, and measured free light chains, in addition to Epstein-Barr virus (EBV) serology, and herpesvirus qPCR from the plasma samples. The presence of anosmia as a proxy for PASC was self-reported at 3–12 months post-COVID. Results: All changes in B cell subpopulation proportions normalized within 200 dsso. Likewise, the acute alterations observed in circulating T follicular helper and T follicular regulatory cell proportions stabilized soon after. Free light chains were high in acute COVID-19, especially in inpatients, but normalized during follow-up. EBV and human herpesvirus 6B (HHV-6B) reactivations were significantly more common in inpatients than outpatients, with reactivation in 47 and 19 % of inpatients and 4.3 and 0 % of outpatients respectively. Anosmia was not significantly associated with any herpesvirus reactivation. Conclusions: The circulating B cell and Th cell subpopulations experience transitional changes during SARS-CoV-2 infection, but these changes recover in follow-up. EBV and HHV-6B reactivations are common in inpatients, but they are not associated with persisting anosmia.Peer reviewe
Review of geochemical processes in CCUS : Mechanisms, processes, and implications
No full textPublisher Copyright: © 2025 International Association for Gondwana ResearchThe emission of greenhouse gases has resulted in the deterioration of the global climate, leading nations worldwide to adopt measures to mitigate the environmental impact of carbon emissions. Carbon dioxide capture, utilization and storage (CCUS) is an emerging large-scale greenhouse gas emission reduction technology with the potential to become an important means of mitigating the greenhouse effect in the future. However, the geochemical reactions accompanying CCUS implementation critically impact system stability and the environment. This review systematically examines the geochemical reaction mechanisms in CCUS reservoirs, including clastic rock, claystone, carbonate rock, igneous rock, and coal, and their effects and risks on CO2 sequestration systems. It has been demonstrated that geochemical reactions are prevalent in various types of CCUS, including mineral dissolution, CO2 mineralization, and adsorption of CO2 by rocks. These reactions may alter rock strength, caprock sealing integrity, fracture development, and groundwater ion concentrations, influencing sequestration outcomes. The study highlights challenges in CCUS geochemical research and proposes future directions. By enhancing understanding of reaction processes and risks, this work provides insights for CCUS operation, monitoring, and research.Peer reviewe
Enhanced carbon monoxide poisoning resistance of Pd/BEA by integrating with CeO2 for low temperature NO adsorption: The role of CeO2 on oxidizing carbon monoxide to CO32−
No full textPublisher Copyright: © 2024Passive NOx adsorption (PNA) represents a promising technology for controlling NOx emissions during cold starts. However, Pd/zeolite catalysts, the latest generation of PNA materials, suffer from serious deactivation under CO concentrations, owing to the CO induced aggregation of highly dispersed Pd sites to PdO particles. In this work, we found that the resistance to CO poisoning of Pd/BEA can be enhanced by integrating with CeO2 through physical mixing. The resulting CeO2 + Pd/BEA composite catalyst demonstrated a notably higher NOx storage capacity (with a NOx/Pd ratio of 1.3) compared to Pd/BEA (0.8). This improvement can be attributed to the interaction between CeO2 and the BEA support, which promoted the formation of nitrates on CeO2 during NO adsorption at 80 °C. More importantly, CeO2 effectively shielded Pd2+ ions from reduction by CO, with 88 % of Pd2+ ions remaining after CO exposure, significantly outperforming Pd/BEA, which retained only 66 %. Spectra characterization and DFT results indicate that the defects in CeO2 strongly trapped CO (−0.973 eV), and the oxidation of trapped CO to CO32− by adsorbed oxygen (Oads) was exothermic (−1.35 eV), which reduced the exposure of Pd2+ ions to CO and inhibited the reduction and agglomeration of Pd2+ ions. While the composite sample still exhibited a slight reduction in NOx storage capacity after CO poisoning, with the NOx/Pd ratio decreasing from 1.3 to 0.9. The DFT simulation results indicate that the Ce-Oads bonds in CeO2 are weakened during the decomposition of the CO32− intermediate. This weakening could potentially facilitate the release of Oads, thereby reducing the Oads content as evidenced by spectral evidence. The decrease of Oads diminished nitrate formation on the CeO2 component in composite sample, resulting in the decline of NOx storage capacity after CO poisoning.Peer reviewe
A Stacking-Based Ensemble Learning Model for Intelligent Ship Trajectory Interpolation
No full textIncomplete ship trajectories caused by irregular Automatic Identification System (AIS) updates pose a critical challenge to reliable modeling of ship behaviors, which underpins the understanding and management of complex maritime traffic systems. Ship trajectory interpolation has therefore become essential for reconstructing missing segments and ensuring data continuity. However, most existing methods adopt an individual interpolation model regardless of varying ship behaviors, which limits their adaptability and may degrade the accuracy of reconstructed trajectories in diverse traffic scenarios. This study presents a Stacking Trajectory Interpolation Model (STIM) that enables adaptive and behavior-aware selection of the most suitable interpolation algorithm for accurate and robust ship trajectory data reconstruction. Specifically, a Markov-based feature extraction approach is first designed to divide trajectories into segments reflective of behavioral patterns, providing informative inputs to support the model’s learning process. Five widely adopted interpolation methods (linear, polynomial, cubic spline, cubic Hermite, and kinematic interpolation) are then implemented in base learners for initial predictions. The meta-learner empowered by a Transformer-based dual-branch multi-classifier subsequently learns the latent relationship between segment features and interpolation performance, enabling the model to support reliable trajectory reconstruction under diverse behavioral patterns. Experimental results using AIS data from Ningbo-Zhoushan Port demonstrate that STIM has strong adaptability and scalability in handling diverse trajectory characteristics, generating more accurate interpolated trajectories compared to conventional methods. Additionally, the impact of different trajectory features on interpolation results is discussed, which further validates the strength of STIM in providing valuable insights for optimizing future trajectory-based application solutions.Peer reviewe