HAL Arts et Métiers
Not a member yet
14127 research outputs found
Sort by
Gamma-induced one-step synthesis of reduced graphene oxide–silver nanoparticles with enhanced properties
No full textInternational audienceThis study presents a novel gamma-induced one-pot synthesis of reduced graphene oxide-silver nanoparticle (rGO-Ag NPs) nanocomposites. Syntheses were conducted in a deoxygenated aqueous medium containing 0.2 g L À1 graphene oxide (GO), silver ions (10 À3 or 10 À2 mol L À1 ), and 0.2 mol L À1 isopropanol at ambient temperature and pressure. Multi-technique characterization confirmed the reduction of GO and silver ions, forming nanocomposites with significantly improved physicochemical and electrochemical properties compared to pristine GO, rGO alone, and rGO-Ag NPs prepared by other methods. UV-Vis absorption spectroscopy revealed tunable optical properties, while UPS measurements provided insights into the energy band structure, highlighting interactions between rGO and Ag NPs that enhance electronic properties. XPS and ATR-FTIR confirmed the successful reduction processes. SEM-EDX analyses demonstrated uniform silver nanoparticle distribution on rGO sheets. The C/O ratio significantly increased after irradiation, with values of 10.8 and 9.6 for composites synthesized with 10 À3 and 10 À2 mol L À1 in silver ions, respectively, compared to 11.2 for rGO alone. Raman spectroscopy showed a lower intensity ratio (I D /I G ) between D and G bands (1.18 for nanocomposites vs.1.40 for rGO), indicating fewer structural defects. Improved thermal stability was evidenced by reduced weight loss (10%) at 300-800 1C. Electrochemical studies revealed exceptional specific capacitance values of 218 F g À1 (10 À3 mol L À1 Ag + at 50 kGy) and 298 F g À1 (10 À2 mol L À1 Ag + at 70 kGy), surpassing the 125.4 F g À1 for rGO alone. These findings highlight the potential of gamma-induced synthesis for producing rGO-Ag NPs nanocomposites for high-performance supercapacitor applications.</div
Designing a Predictive Super-Twisting Sliding Mode Control for Floating Offshore Wind Turbines in Region 2
No full textInternational audienceThis study focuses on designing an advanced control strategy for floating offshore wind turbines (FOWTs) to maximize power generation for wind speeds between 3 and 11.25 m/s (region 2) while reducing fatigue loads on the system. One of the main challenges in designing a control law for FOWTs is their nonlinearity and high degree of freedom. To address this issue, various robust nonlinear control methods with reduced knowledge of the model, such as super-twisting controller (STW), and model-based control approaches, such as model predictive control (MPC), are used. However, both of these controllers suffer from disadvantages, such as slow transient time for STW and a lack of robustness against uncertainties for MPC. To address this problem, this paper integrates optimal predictive control and STW to overcome their deficiencies and achieve superior performance. Simulation studies is conducted on the 5MW OC4 FOWT, which is modeled by OpenFAST. The comparison results with the ROSCO method, which is the Reference Open Source Controller for Wind Turbines, show that the proposed method improves power generation with a reduction of pitch rate. The primary contribution of this paper can be summarized as: 1) Development of a control law based on optimal predictive control and STW for a FOWT to achieve optimized performance and robustness, 2) Evaluation of the performance of predictive super-twisting sliding mode (PSTW) versus the baseline controller (ROSCO) with quantitative indicators
Proposition d’une approche méthodologique outillée pour la gestion des connaissances dans les processus de conception de produits : application à la fabrication additive
No full textIndustry 4.0 requires effective knowledge capitalization to optimize design processes within manufacturing and digital services companies (DSCs). In a context marked by the growing complexity of industrial environments and the high turnover of consultants, companies face the challenge of effectively transferring knowledge from experts to novice designers. To address this issue, this thesis proposes KARMEN (Knowledge Access for Request Manufacturing and Engineering by Network graph), a knowledge graph based on a method consisting of four steps: FAME (Find, Acquire, Model, Exchange). These steps enable knowledge to be found, acquired and exploited.This thesis then addresses our following research question: "How to capture heterogeneous data and the knowledge of expert designers within an ESN to assist novices in the design process?"To do this, we propose the PPR-FBS-8M data model, which offers a holistic view of the product, process and resources. This model makes it possible to integrate and organize heterogeneous data, whether structured (such as data records) or unstructured (such as expert reports), thus facilitating the transmission of knowledge between experts and novices. It should also be noted that our data model is based on the fusion of a knowledge graph and the double diamond design process.This work has also given rise to two scientific publications. The first, entitled "KARMEN: Redefining collaboration and expertise sharing through an innovative knowledge graph framework: a case study in additive manufacturing," was presented at the PLM 23 conference. The second publication, "KARMEN: A Knowledge Graph Based Proposal to Capture Expert Designer Experience and Foster Expertise Transfer," was published in the peer-reviewed journal IJIDEM. These contributions attest to the significant impact of KARMEN on the design and knowledge sharing process.Finally, in line with our method, we have developed an educational support within the LCPI laboratory, intended to raise awareness of our approach among future users. This support offers a practical immersion through the manipulation of tangible Intermediate Representations, thus promoting a better understanding of the various opportunities offered by manufacturing processes, whether additive, formative or subtractive.This educational approach aims not only to strengthen access to technical knowledge, but also to facilitate the integration of new designers into the design and prototyping processes. Furthermore, it opens up promising prospects for the continuing education and professional development of designers, enabling them to acquire essential skills and adapt to the rapid developments of Industry 4.0.L’Industrie 4.0 exige une capitalisation efficace des connaissances pour optimiser les processus de conception au sein des entreprises manufacturières et des entreprises de services numériques (ESN). Dans un contexte marqué par la complexité croissante des environnements industriels et la rotation élevée des consultants, les entreprises doivent relever le défi de transférer efficacement les savoirs des experts vers les concepteurs novices. Pour répondre à cette problématique, cette thèse propose KARMEN (Knowledge Access for Request Manufacturing and Engineering by Network graph), un graphe de connaissances qui repose sur une méthode composée de quatre étapes : FAME (Find, Acquire, Model, Exchange). Ces étapes permettent de trouver, d’acquérir et d’exploiter la connaissance.Cette thèse aborde alors notre question de recherche suivante : « Comment capter les données hétérogènes et la connaissance des concepteurs experts au sein d'une ESN pour assister les novices dans le processus de conception ? »Pour cela, nous proposons le modèle de données PPR-FBS-8M, qui offre une vision holistique du produit, du processus et des ressources. Ce modèle permet d’intégrer et d’organiser des données hétérogènes, qu’elles soient structurées (comme des relevés de données) ou non structurées (telles que des rapports d’expertise), facilitant ainsi la transmission des savoirs entre experts et novices. Soulignons, également que notre modèle de données repose sur la fusion d’un graphe de connaissances et du processus de conception en double diamant.Ces travaux ont également donné lieu à deux publications scientifiques. La première, intitulée "KARMEN: Redefining collaboration and expertise sharing through an innovative knowledge graph framework: a case study in additive manufacturing," a été présentée lors de la conférence PLM 23. La seconde publication, "KARMEN: A Knowledge Graph Based Proposal to Capture Expert Designer Experience and Foster Expertise Transfer," a été publiée dans le journal à comité derelecture IJIDEM. Ces contributions attestent de l'impact significatif de KARMEN sur leprocessus de conception et de partage des connaissances.Enfin, dans la continuité de notre méthode, nous avons élaboré un support pédagogique au sein même du laboratoire LCPI, destiné à sensibiliser les futurs utilisateurs à notre approche. Ce support offre une immersion pratique à travers la manipulation de Représentations Intermédiaires tangibles, favorisant ainsi une meilleure compréhension des diverses opportunités offertes par les procédés de fabrication, qu'ils soient additifs, formatifs ou soustractifs.Cette approche pédagogique vise non seulement à renforcer l'accès aux connaissancestechniques, mais également à faciliter l'intégration des nouveaux concepteurs dans lesprocessus de conception et de prototypage. De plus, elle ouvre des perspectives prometteuses pour la formation continue et le développement professionnel des concepteurs, en leur permettant d'acquérir des compétences essentielles et de s'adapter aux évolutions rapides de l'Industrie 4.0
Prevailing effect of residual stresses and defects on the fatigue strength of net-shape parts produced with Laser Powder Bed Fusion (L-PBF) 316L stainless steel
No full textInternational audienceLaser Powder Bed Fusion (L-PBF) additive manufacturing enables the production of complex-shaped parts with high mechanical resistance. Such components exhibit a multi-scale microstructure, internal, sub-surface, and surface defects, a rough surface finish and a residual stress gradient in the as-built net-shape condition. All of these factors can alter the fatigue behaviour. This study aims to improve the understanding of the combined effect of various surface parameters on the fatigue behaviour of L-PBF 316L stainless steel by conducting an extensive experimental campaign. Uni-axial fatigue tests were carried out on six batches having an as-built or heat-treated microstructure and a net-shape, pre-corroded net-shape, single-defect net-shape or polished surface condition. Results were compared with data from the literature: as-built polished, pre-corroded polished, or single-defect polished specimens. Studied defects were process-induced (e.g. lack of fusion, spatter, gas pore) and artificial (i.e. corrosion pit, electric discharge machined defect). Their sizes ranged from 10 to 700 μm. The residual stresses gradients were characterized by X-ray diffraction. A Kitagawa–Takahashi diagram was used to illustrate the effects of the various parameters on the fatigue behaviour. Residual stresses and defects were the most influential factors on the fatigue strength of net-shape specimens over surface condition and sub-surface microstructure
Recurrent Neural Networks model for injury prevention within a professional rugby union club: a proof of concept over one season
No full textInternational audienceBackgroundIn professional rugby, injury prevention and player availability are major challenges. Sports analytics use data from trainings and matches to address these issues. This study leveraged comprehensive daily data from a professional rugby club to predict players' readiness for training. Using this metric helped assess its effectiveness in predicting intrinsic injuries and improving injury prevention strategies.MethodsModels including logistic regression, decision trees, and Long Short-Term Memory-based neural networks, were evaluated for their predictive accuracy and ability to discern patterns indicative of injury risks or readiness for physical activities.FindingsThe study demonstrated that long-short term memory and convolutional one-dimension models outperform traditional machine learning methods in analyzing players' physical conditions. This approach may support earlier identification of injury risks and inform workload management. Using model evaluation and interpretability techniques, including Local Interpretable Model-Agnostic Explanations (LIME) module, the study provided a framework for sports scientists, coaches, and medical staff to mitigate injury risks and optimize training sessions.InterpretationAs a preliminary exploration, this study paves the way for further research into the integration of machine learning and neural networks in sports science, promising transformative impacts on injury prevention strategies in rugby
Data driven modelling as a new route to design PLA based materials with improved barrier properties
No full textInternational audienc
Underwater Netting Detection and Distance Estimation using Optic Flow
No full textObstacle detection and avoidance remains a major challenge for automatic underwater navigation. While there exists abundant literature on this topic, certain types of poorly defined obstacles are difficult to detect, such as underwater netting. Optic Flow is a bio-inspired optical cue widely used in land-based autonomous navigation and by autonomous aircraft. Requiring only small sensors or a monocular camera, it is a minimalist but reliable source of information. However, because of the low visibility, optical distortions and overall poor quality of optical measurements underwater, examples of underwater applications of this technique remains scarce.This study proposes an underwater netting detection algorithm called uNOWA (underwater Netting Optical floW detection Algorithm), adapted to the specifics of underwater vision. This method also estimate the distance between the camera and the net, by combining the translational optic flow with the velocity. A number of experiments have been carried out under real-life underwater conditions using different nets. The results show an effective net detection and give an estimation of the distance to the detected netting.</div
Simuler les interactions humain-IA : plans de recherche autour de l'émergence d'une décision hybride
No full textInternational audienceDe nombreux secteurs intègrent de nouvelles briques d'IA dans leurs activités. Ces programmes posent des problèmes de fiabilité, de compréhension et donc de contrôle. Malgré l'aspiration de garder l'humain dans la boucle, les implications d'une telle hybridation des activités sont encore peu connues. Dans cet article, nous proposons des cadres de modélisation afin de cartographier la littérature sur les interactions humain-IA, et de planifier nos recherches à venir avec la Marine, intégrant expérimentation humaine et simulation numérique.1. L'utilisateur n'a pas a priori de moyen de distinguer le fruit d'une programmation humaine de celui d'une optimisation statistique.</div
Flying in air ducts
No full textAbstract Air ducts are integral to modern buildings but are challenging to access for inspection. Small quadrotor drones offer a potential solution, as they can navigate both horizontal and vertical sections and smoothly fly over debris. However, hovering inside air ducts is problematic due to the airflow generated by the rotors, which recirculates inside the duct and destabilizes the drone. In this article, we map the aerodynamic forces that affect a hovering drone in a duct using a robotic setup and a force/torque sensor. Based on the collected aerodynamic data, we identify a recommended position for stable flight, which is not the center of a circular duct. We then develop a neural network-based positioning system that leverages low-cost time-of-flight sensors. By combining these aerodynamic insights and the data-driven positioning system, we show how to improve the stability of a small quadrotor drone (here, 180 mm) inside small air ducts (down to 350 mm diameter) and fly autonomously over 2 m
Flying in air ducts
No full textInternational audienceAir ducts are integral to modern buildings but are challenging to access for inspection. Small quadrotor drones offer a potential solution, as they can navigate both horizontal and vertical sections and smoothly fly over debris. However, hovering inside air ducts is problematic due to the airflow generated by the rotors, which recirculates inside the duct and destabilizes the drone, whereas hovering is a key feature for many inspection missions. In this article, we map the aerodynamic forces that affect a hovering drone in a duct using a robotic setup and a force/torque sensor. Based on the collected aerodynamic data, we identify a recommended position for stable flight, which corresponds to the bottom third for a circular duct. We then develop a neural network-based positioning system that leverages low-cost time-offlight sensors. By combining these aerodynamic insights and the data-driven positioning system, we show that a small quadrotor drone (here, 180 mm) can hover and fly inside small air ducts, starting with a diameter of 350 mm. These results open a new and promising application domain for drones