HAL-Artois
Not a member yet
27242 research outputs found
Sort by
Une analyse en composantes principales pour identifier les stratégies utilisées par les étudiant-e-s en réussite lors d'un apprentissage “A son Rythme” pour un enseignement sur les incertitudes
No full textInternational audienceAs part of the “at its own pace” learning program which has been running at Le Mans Université for over 10 years, students have the opportunity to make a number of choices within the framework of their course of uncertainties: choice of resources (handouts, videos, etc.), their pace of work, their working time outside class sessions, and the times at which they attempt to validate the achievement of learning objectives. These items are traces of what we call the “strategy”. A collection of anonymized data at the end of each session, followed by a principal components analysis, enables us to identify the strategies that are statistically the most successful. A presentation of the issues at stake and the results can then be given to students at the start of the course, so that they can choose these strategies with full knowledge of the facts. As each strategy parameter can be quantified, the approach allows to move away from a moralistic discourse (“You have to work harder.”) to propose a scientifically-measured effort that makes sense. The analysis presented at concerns two successive years of the same course, and highlights the robustness of the parameters that make up these effective learning strategies.Lors d’un apprentissage “A son Rythme” qui est expérimenté depuis plus de 10 ans ausein de Le Mans Université, les étudiant-e-s ont la possibilité de faire un certain nombre de choix dansle cadre de leur cours d’incertitudes : choix des ressources (polycopié, vidéos...), leur rythme de travail,leur temps de travail hors séances de cours, les moments auxquels ils ou elles tentent de valider l’atteinte des objectifs d’apprentissage. Ces items sont des traces de ce que nous appelons la “stratégie”. Un recueil de données anonymisées à la fin de chaque séance, puis une analyse en composantes principales,permettent d’identifier les stratégies statistiquement les plus performantes. Une présentation des enjeuxet des résultats peut alors être faite aux étudiant-e-s en début d’enseignement afin qu’ils choisissent cesstratégies en toute connaissance de cause. Chaque paramètre de la stratégie pouvant être quantifié, ladémarche permet de quitter un discours moralisateur (“Il faut travailler plus.”) pour proposer un effort mesuré scientifiquement et qui fait sens. L’analyse présentée concerne deux années successives du même cours, et met en évidence la robustesse des paramètres qui composent ces stratégies d’apprentissage efficaces
Machine Learning-Based Detection of Inter-Turn Short-Circuit Faults in Synchronous Machines Using Stray Flux Time-Domain Analysis
No full textInternational audienceThis paper deals with an inter-turn short-circuit fault detection approach for synchronous machines using timedomain analysis of the external magnetic field. We compare single-sensor and dual-sensor configurations, extracting 22 timeseries features using the Catch22 algorithm to classify healthy and faulty states. Machine learning classifiers are employed; they are initially trained on simulated data and later refined using a mixed dataset incorporating experimental signals to improve generalization. Results show that logistic regression struggles with nonlinear patterns, while random forest demonstrates superior robustness. The dual-sensor configuration further enhances detection, confirming it as the most effective strategy for accurate and reliable fault classification
Improving Water Safety and Long-term Engagement in Aquatic Physical Activity: The Aquatic Literacy Framework
No full textInternational audienceYoung people worldwide are affected by two main health-related issues: physical inactivity and drowning accidents. Encouraging participation in aquatic physical activities in a safe way could address this dual concern. Aligning theoretical frameworks of physical literacy and water competencies could empower young people with key skills to engage safely and sustainably in aquatic environments. The aim of this theoretical paper is to provide a philosophical, ontological, and epistemological rationale for merging these two frameworks into a new concept: aquatic literacy. The multidimensional concept is presented and describes the intertwined physical, psychosocial, and cognitive domains, which provide a means to empower young people to safely enjoy aquatic opportunities and achieve individual health benefits. This new concept needs supportive empirical data, such as the level of individuals in each domain of aquatic literacy, to be a strong framework for guiding aquatic educators and stakeholders to better engage and protect people worldwide.Les jeunes à travers le monde peuvent être confrontés à deux enjeux sanitaire : l’inactivité physique et les accidents de noyade. Encourager la participation à des activités physiques aquatiques de manière sécurisée pourrait répondre à cette double problématique. L’alignement des cadres théoriques de la Littératie Physique et de Compétences Aquatiques pourrait permettre aux jeunes d’acquérir les compétences clés nécessaires pour évoluer en toute sécurité et de manière durable dans les environnements aquatiques. L’objectif de cet article théorique est de fournir une justification philosophique, ontologique et épistémologique en faveur de la fusion de ces deux cadres en un nouveau concept : la Littératie Aquatique. Ce concept multidimensionnel est présenté en décrivant les domaines physiques, psychosociaux et cognitifs interconnectés, qui offrent un moyen d’éduquer les jeunes afin qu’ils puissent profiter en toute sécurité des opportunités aquatiques et en retirer des bénéfices pour leur santé. Ce concept nécessite des données empiriques à l’appui, telles que le niveau des individus dans chacun des domaines de la littératie aquatique, afin de constituer un cadre solide permettant aux éducateurs aquatiques et aux parties prenantes de mieux engager et protéger les populations à l’échelle mondiale
L’intelligence artificielle au service de la didactique du Français sur Objectif Universitaire : utopie concrète ou miroir aux alouettes ?
No full textInternational audienc
Outils linguistiques, discours universitaires, dispositifs didactiques : pistes d'ingénierie(s) pédagogique(s) pour l'accompagnement des étudiants
No full textInternational audienc
Les Cifre en tiers-lieux: interstices singuliers entre recherche et action
No full textNational audienceSelon le rapport France Tiers-Lieux (2021), les tiers-lieux font émerger des pratiques de recherche et d’innovation en dehors des cadres habituels de la science (laboratoires de recherche, grandes entreprises). Ces espaces d’expérimentation invitent à penser les façons de faire recherche autrement par rapport à larecherche académique classique. Toutefois, la littérature scientifique comporte peu de travaux portant spécifiquement sur les interactions qu’entretiennent les tiers-lieux et la recherche.Après avoir situé les approches de recherche en tiers-lieux, nous proposons d’en saisir les manifestations et les enjeux, à travers quatre récits d’expérience en contrat doctoral Cifre, présentant des situations de recherche singulières à la rencontre entre savoirs d’action et connaissances scientifiques
Driving the future of cosmetics, fragrances and foods with COSMO-RS.Part 2-From theory to practice
No full textInternational audienceThe second part of this two-part review provides a comprehensive overview of the practical applications of COSMO-RS methodology in the formulation of cosmetics, fragrances, and foods. The current literature predominantly focuses on the recovery and characterization of ingredients, with fewer studies addressing end-use formulations. Four major types of applications have been identified and are discussed (i) the extraction of active ingredients and additives from biomass, by far the most documented, (ii) their solubilization in skin-compatible or ingestible solvents, (iii) the selection of fragrant molecules to establish a predefined olfactory profile, and finally, (iv) the physico-chemical characterization of ingredients to maximize their sensory performance and safety. For further familiarization with COSMO-RS, the typical calculation procedures are demonstrated step-by-step for key target properties
SAT-Based CEGAR Method for the Hamiltonian Cycle Problem Enhanced by Cut-Set Constraints
No full textInternational audienceIn this paper, we propose an enhancement to the SAT-based counterexample-guided abstraction refinement (CEGAR) approach for solving the Hamiltonian Cycle Problem (HCP). Many SAT-based methods for HCP have been proposed, including a CEGAR-based method that repeatedly solves a relaxed version of HCP strengthened by counterexamples. However, when the counterexample space - represented by the full set of subcycle partitions - is large, it becomes difficult to find a solution. To address this, we introduce cut-set constraints in the refinement step, replacing traditional subcycle blocking constraints. Our evaluation shows that these cut-set constraints achieve equal or better reduction in the counterexample space, making it easier to find valid solutions. We further assessed performance using all 1001 instances from the FHCP challenge set and confirmed that the proposed method solved 937 instances within 1800 seconds, outperforming both the existing eager and CEGAR encodings (which solved at most 666 instances). This demonstrates the effectiveness of incorporating cut-set constraints into SAT-based CEGAR approaches
Continuous epoxidation of used cooking oils using an automated slug-flow millireactor
No full textInternational audienceThis study describes the development of an automated experimental platform designed for the continuousepoxidation of used cooking oils (UCOs) in slug-flow millireactors. This system transforms UCOs into high-value,second-generation oleochemicals, employing an intensified process, ensuring reproducibility, high yields, andenhanced productivity. The epoxidation was conducted via Prilezhaev reaction using H2O2 as an oxidizing agent,peracetic acid as an oxygen carrier, and H2SO4 as a catalyst. Different vegetable oils were tested to assess theimpact of unsaturation content and oil properties on the process performance, and it was found that viscosity hada high effect on the hydrodynamic patterns within the reactor and that specific operating conditions wererequired to reach slug flows with each feedstock. Then, preliminary experiments with UCO yielded suitableoperating conditions to ensure a proper slug flow regime. It was found that the high content of polar compoundsin UCO had a significant impact in the hydrodynamics of the reactor because those components induce coa-lescence with the aqueous phase. Thus, the levels of polar components and moisture in UCO can indicate itssuitability for further epoxidation in the slug-flow reactor and the necessity for pretreatment. Subsequently, anexperimental simplex evolutionary optimization was deployed to verify a selectivity towards oxirane groups >80 %, with conversions up to 86 % and yields up to 73 %. The optimal operating conditions were 77.4 ◦C, anH2O2 to oil ratio of 0.84:1, an acid to oil ratio of 0.32:1, and a residence time of 22.7 min. Under these conditions,a conversion of 82 %, selectivity of 86 %, and productivity of 0.75 kg OO⋅m 3⋅min 1 were achieved, and thecorresponding epoxidized UCO had an oxirane oxygen content of 4.02 wt%
