Archive ouverte de Centrale Lyon
Not a member yet
32420 research outputs found
Sort by
Développement d'une méthode unifiée de caractérisation de l'efficacité des fluides de coupe
No full textThis thesis introduces a unified method for characterizing the impact of cutting fluids in machining, based on a tribological approach tailored to industrial needs. The methodology relies on a combination of standard tests, representative tests, and numerical simulations. These tests are used to assess lubrication regimes, fluid persistence at the contact interface, and tool wear.Standard tests are conducted using a block-on-ring tribometer, with various materials and controlled test conditions. They allow the plotting of Stribeck curves and the identification of lubrication regimes for each fluid studied. Representative tests are carried out on a dedicated tribometer mounted on a machine tool, in order to simulate contact conditions closer to those encountered in machining, particularly in terms of pressure and relative speed. This setup makes it possible to observe the fluid's ability to enter the contact zone and then remain within it.Experimental validation is then performed through machining tests, with measurement of insert wear. All collected data are integrated into a tribological performance indicator, which combines wear, persistence, and lubrication regime parameters. This indicator provides an objective ranking of the tested fluids. The developed method is designed to be adaptable to various machining configurations and fluid families, with potential extensions toward soluble oils, which are not addressed in this manuscript.Keywords: Tribology, Cutting fluids, Lubrication regime, Contact persistenceCette thèse propose une méthode unifiée pour caractériser l'impact des fluides de coupe en usinage, en s'appuyant sur une approche tribologique pensée pour répondre aux besoins industriels. Elle repose sur une combinaison d'essais standards, d'essais représentatifs et de simulations numériques. Ces essais permettent d'évaluer les régimes de lubrification, la persistance des fluides au contact et l'usure des outils.Les essais sur standards sont réalisés sur un tribomètre block-on-ring, avec différents matériaux et des conditions d'essai contrôlées. Ils permettent de tracer les courbes de Stribeck et d'identifier les les régimes de lubrification de chacun des fluides de l'étude. Les essais représentatifs sont menés sur un tribomètre dédié, monté sur machine-outil, afin de simuler des conditions de contact plus proches de celles rencontrées en usinage, notamment en termes de pression et de vitesse relative. Cette configuration permet d'observer la capacité du fluide à pénétrer le contact, puis à s'y maintenir.Une validation expérimentale est ensuite réalisé à l'aide d'essais d'usinage, avec mesure de l'usure des plaquettes. L'ensemble des données obtenues est intégré dans un indicateur de performance tribologique, qui combine les paramètres d'usure, de persistance et de régimes de lubrification des fluides. Ce dernier permet de proposer un classement objectif des fluides testés. La méthode développée est conçue pour être adaptable à différentes configurations d'usinage et à diverses familles de fluides, avec des perspectives d'extension vers les huiles solubles, qui ne sont pas abordées dans ce manuscrit.Mots-clés: Tribologie, Fluides de coupe, Régime de lubrification, Persistance au contac
Optimizing Perpendicular Flux in PCB Embedded Inductors via Shift-Angle Topologies under DC Excitation
No full textInternational audienceThe integration of magnetic components into printed circuit boards (PCBs) presents a promising solution for enhancing electromagnetic performance while minimizing space and manufacturing complexity in power electronics. A key challenge in this domain is the angular misalignment of inductor turns relative to the magnetic core centerline-often inevitable due to PCB manufacturing constraints. This misalignment distorts the magnetic field, increasing the perpendicular (Z-direction) magnetic flux density through copper traces and leading to elevated eddy current losses, leading to elevated eddy current losses in AC applications. This study investigates various angle alignment topologies inPCB-embedded solenoid inductors to mitigate electromagnetic losses and optimize inductive performance. To isolate the influence of angular alignment from frequency-dependent phenomena like skin effect, proximity effect, and core losses, simulations were conducted under DC conditions using Ansys Maxwell. Several angular configurations were analyzed to evaluate their effects on flux distribution and eddy current formation. Results indicate that altering the angle alignment significantly impacts flux behavior, with Configuration 2 (A = -Q), featuring two opposite angles symmetrically placed at the midpoint of the turns, demonstrating the lowest flux loss and eddy current dissipation. Experimental validation confirmed the simulation findings, showing electromagnetic losses remained within a 6% margin.This study confirms that optimizing turn alignment can substantially improve the efficiency of PCB-integrated inductors. Future work will explore further enhancements through advanced winding techniques (e.g., litz wire), alternative geometries (e.g., toroidal shapes), and the integration of active components for high-frequency, highefficiency converter applications
Comment les réarrangements chromosomiques façonnent les génomes : étude par modélisation et simulations
No full textThe origins of genome complexity, as well as the determinants of genome size, remain widely debated. This thesis shows that chromosomal rearrangements are a key factor in the evolution of genome architecture in terms of size and complexity. In particular, it shows that genome size and coding fraction are closely linked to the selection for robustness to chromosomal rearrangements, which is notably modulated by population size and mutation rate. We first studied the impact of chromosomal rearrangements on the evolution of bacterial genome architecture. To this end, the thesis relies on computer simulations and mathematical modeling. In particular, for the simulations, it relies on Aevol, a software designed to study prokaryote genome structure evolution, which allows for chromosomal rearrangements to act directly on the genomic sequence of individuals. Using Aevol, we were able to show that chromosomal rearrangements are essential for sustaining long-term adaptation, but also for stabilizing genome size. This result enabled us to show, through large-scale simulation campaigns, that the pressure imposed by rearrangements on genome size is modulated by both mutation rate (which modifies genome robustness) and population size (which modifies the efficiency of selection for robustness). This result was then confirmed by a mathematical model showing how these two parameters determine an equilibrium non-coding genome proportion.The second part of the thesis focuses on generalizing the previous results to eukaryotic genomes. First, it presents a new version of Aevol developed specifically for the project that entails diploid organisms with linear chromosomes that reproduce sexually and undergo a mandatory meiotic recombination event. Using this model, we show that eukaryote-like genomes react to changes in mutation rate and population size in the same way as prokaryote-like genomes. In the last chapter, we show that the reproductive mode is also an important determinant of genome architecture, as self-fertilization leads to more streamlined genomes. To conclude, this PhD thesis presents a new globally coherent and self-contained framework for understanding fundamental aspects of genome size evolution, focused on the direct and indirect impact of mutations, especially chromosomal rearrangements, and how they affect the future of each lineage. We also show how other parameters, such as the population size and the reproduction mode (asexual, sexual, self-fertilization), interact with these mutations and modulate their impact on genome size evolution. Taken together, these results contribute to a unifying view of the evolution of genome architecture and complexity along the tree of life.Les origines de la complexité des génomes, ainsi que les déterminants de la taille des génomes, restent largement débattus. Cette thèse montre que les réarrangements chromosomiques sont un facteur clé de l'évolution de l'architecture du génome en termes de taille et de complexité. Elle montre en particulier que la taille et la fraction codante des génomes sont étroitement liées à la sélection de la robustesse aux réarrangements chromosomiques, et que celle-ci est notamment modulée par la taille de la population et le taux de mutation. Dans un premier temps, nous avons étudié l'impact des réarrangements chromosomiques sur l'évolution de l'architecture des génomes bactériens. Pour cela, la thèse s'appuie sur des simulations informatiques et des modélisations mathématiques. En particulier, pour les simulations, elle s'appuie sur Aevol, un logiciel conçu pour étudier l'évolution de la structure des génomes procaryotes, qui permet aux réarrangements chromosomiques d'agir directement sur la séquence génomique des individus. En utilisant Aevol, nous avons pu montrer que les réarrangements chromosomiques sont essentiels pour soutenir l'adaptation à long terme, mais aussi pour stabiliser la taille du génome. Ce résultat nous a permis de montrer, par des campagnes de simulation à grande échelle, que la pression imposée par les réarrangements sur la taille du génome est modulée à la fois par le taux de mutation (qui modifie la robustesse des génomes) et par la taille de la population (qui modifie l'efficacité de la sélection pour la robustesse). Ce résultat a ensuite été confirmé par un modèle mathématique qui met en évidence comment ces deux paramètres déterminent une proportion d'équilibre du génome non codant. La deuxième partie de la thèse se concentre sur la généralisation des résultats précédents aux génomes eucaryotes. Tout d'abord, elle présente une nouvelle version d'Aevol développée spécifiquement pour le projet, qui modélise des organismes diploïdes avec des chromosomes linéaires se reproduisant sexuellement et subissant un événement de recombinaison méiotique obligatoire. Avec ce modèle, nous montrons que les génomes de type eucaryote réagissent aux changements du taux de mutation et de la taille de la population de la même manière que les génomes de type procaryote. Dans le dernier chapitre, nous montrons que le mode de reproduction est également un déterminant important de l'architecture du génome, car l'auto-fécondation conduit à des génomes réduits.En conclusion, cette thèse de doctorat présente un nouveau cadre permettant de comprendre les aspects fondamentaux de l'évolution de la taille des génomes, en se concentrant sur l'impact direct et indirect des mutations, et en particulier des réarrangements chromosomiques, et sur la manière dont elles affectent l'avenir de chaque lignée. Nous montrons également comment d'autres paramètres, tels que la taille de la population et le mode de reproduction (asexué, sexué, autofécondation), interagissent avec ces mutations et modulent leur impact sur l'évolution de la taille du génome. L'ensemble de ces résultats contribue à une vision unifiée de l'évolution de l'architecture et de la complexité des génomes le long de l'arbre de la vie
Approximate mean curvature flows of a general varifold, and their limit spacetime Brakke flow
No full textWe propose a construction of mean curvature flows by approximation for very general initial data, in the spirit of the works of Brakke and of Kim & Tonegawa based on the theory of varifolds. Given a general varifold, we construct by iterated push-forwards an approximate time-discrete mean curvature flow depending on both a given time step and an approximation parameter. We show that, as the time step tends to , this time-discrete flow converges to a unique limit flow, which we call the approximate mean curvature flow. An interesting feature of our approach is its generality, as it provides an approximate notion of mean curvature flow for very general structures of any dimension and codimension, ranging from continuous surfaces to discrete point clouds. We prove that our approximate mean curvature flow satisfies several properties: stability, uniqueness, Brakke-type equality, mass decay. By coupling this approximate flow with the canonical time measure, we prove convergence, as the approximation parameter tends to , to a spacetime limit measure whose generalized mean curvature is bounded. Under an additional rectifiability assumption, we further prove that this limit measure is a spacetime Brakke flow
A local-global principle for twisted flag varieties: We prove a local-global principle for twisted flag varieties over a semiglobal field.
No full textInternational audienceWe prove a local-global principle for twisted flag varieties over a semiglobal field
A Li-Yau and Aronson-Bénilan approach for the Keller-Segel system with critical exponent
No full textWe prove Li--Yau and Aronson--Bénilan type estimates for the parabolic-elliptic Keller-Segel system with critical exponent m=2-\f 2d, i.e. lower bounds on the Laplacian of a suitable notion of pressure in any dimension. We show that these estimates entail bounds on the density, depending on its initial mass, up to the critical mass case for . We deduce from these results the global existence of smooth solutions in two cases: first, when initial data is merely a measure but has sufficiently small mass; and second, when the initial free energy is bounded, and the mass is subcritical or critical. Our argument requires a careful study of the subsolutions of the Liouville and Lane-Emden equations arising in the model
U(h)-finitely generated g-modules and coherent families
No full textThis thesis investigates the relationship between two categories of representations over a finite-dimensional simple Lie algebra g: the category Wadm of admissible weight modules, consisting of g-modules on which a fixed Cartan subalgebra h ⊆ g acts semi-simply with weight spaces of uniformly bounded dimension; and the category A of U(h)-finite g-modules, consisting of modules finitely generated over U(h), the universal enveloping algebra of h. We establish that both categories Wadm and A have objects with Gelfand-Kirillov dimension bounded by the rank of g. As a consequence of this result, we prove that A shares several structural properties with Wadm, such as the finite length of its objects, a block decomposition corresponding to generalized central characters and the existence of infinite dimensional objects only when g is of type A or C. The weighting functor W provides another link between these categories, mapping a U(h)- finite modules to admissible weight modules. By studying W and its left derived functors, we transfer properties of simple admissible weight modules to A. In particular, we prove that infinite dimensional simple U(h)-modules are U(h)-torsion free and locally free at all but finitely many maximal ideals of U(h). We introduce the concept of almost equivalence of weight modules and show that, if M ∈ A is simple of infinite dimension, then W(M) is almost equivalent to a finite number of copies, called coherent-multiplicity, of a unique irreducible semi-simple coherent family - a class of modules introduced and classified by Mathieu in [Mat00]. Using this framework, we classify simple U(h)-finite modules with coherent-multiplicity one, except for those with regular central character when g = sl(n + 1) for n ≥ 3. Finally, for each positive integer m, we construct a family of simple sl(n + 1)-modules with coherent-multiplicity m. Moreover, we show that any U(h)-finite module with coherentmultiplicity greater than one cannot be a weight module with respect to any Cartan subalgebra h′ ⊆ g.Cette thèse étudie la relation entre deux catégories de représentations d’une algèbre de Lie simple de dimension finie g : la catégorie Wadm des modules de poids admissibles, constituée de g-modules sur lesquels une sous-algèbre de Cartan fixe h ⊆ g agit semi-simplement avec des espaces de poids de dimension uniformément bornée ; et la catégorie A des g-modules U(h)-finis, constituée de modules de type fini sur U(h), l’algèbre enveloppante universelle de h. Nous établissons que les objets des deux catégories Wadm et A ont une dimension de Gelfand-Kirillov bornée par le rang de g. En conséquence, nous prouvons que A partage plusieurs propriétés structurelles avec Wadm, telles que la longueur finie de ses objets, une décomposition en blocs correspondant aux caractères centraux généralisés, et l’existence d’objets de dimension infinie uniquement lorsque g est de type A ou C. Le foncteur de pondération W, qui associe à un module U(h)-fini un module de poids admissible, fournit un autre lien entre ces catégories. En étudiant W et ses foncteurs dérivés à gauche, nous transférons des propriétés des modules de poids admissibles simples à A. En particulier, nous prouvons que les modules U(h)-finis simples de dimension infinie sont sans torsion sur U(h) et localement libres en tout idéal maximal de U(h), sauf pour un nombre fini d’entre eux. Nous introduisons la notion de presque équivalence pour les modules de poids et montrons que, si M ∈ A est simple de dimension infinie, alors W(M) est presque équivalent à un nombre fini de copies, appelé la multiplicité-cohérente, d’une unique famille cohérente semi-simple irréductible — une classe de modules introduite et classifiée par Mathieu dans [Mat00]. En utilisant ce cadre, nous classifions les modules U(h)-finis simples de cohérente multiplicité un, à l’exception de ceux ayant un caractère central régulier lorsque g = sl(n+1) pour n ≥ 3. Enfin, pour chaque entier positif m, nous construisons une famille de sl(n + 1)-modules simples de multiplicité-cohérente m. De plus, nous montrons que tout module U(h)-fini avec une multiplicité-cohérente supérieure à un ne peut pas être un module de poids par rapport à aucune sous-algèbre de Cartan h′ ⊆ g
Breast Cancer Diagnosis With Explainable Artificial Intelligence (XAI): Uncovering Strengths and Biases
No full textInternational audienceBreast cancer is one of the most common malignancies afflicting women globally, necessitating the use of cutting-edge AI approaches in diagnostic procedures to improve patient outcomes drastically. However, one key difficulty with the most recent AI models is a lack of transparency, making it difficult for medical practitioners to implement these technologies to increase diagnostic accuracy. Many explainable AI (XAI) solutions have been developed to overcome this issue. This survey focuses on applying XAI approaches in breast cancer detection and diagnosis, particularly emphasizing their role in increasing model transparency and clinical decision-making. The article also provides insights into the inherent biases in the most recent machine learning models towards specific XAI approaches, such as the compatibility of Convolutional Neural Networks (CNNs) with visual explanation methods and tree-based models with feature significance evaluations. Finally, the article covers the obstacles to using XAI technologies in clinical practice and the significance of defining standard measures for assessing their performance
Beyond Batch Learning: Global Awareness Enhanced Domain Adaptation
No full textInternational audienceIn domain adaptation (DA), the effectiveness of deep learning-based models is often constrained by batch learning strategies that fail to fully apprehend the global statistical and geometric characteristics of data distributions. Addressing this gap, we introduce "Global Awareness Enhanced Domain Adaptation" (GAN-DA), a novel approach that transcends traditional batch-based limitations. GAN-DA integrates a unique predefined feature representation (PFR) to facilitate the alignment of cross-domain distributions, thereby achieving a comprehensive global statistical awareness. This representation is innovatively expanded to encompass orthogonal and common feature aspects, which enhances the unification of global manifold structures and refines decision boundaries for more effective DA. Our extensive experiments, encompassing 27 diverse cross-domain image classification tasks, demonstrate GAN-DA's remarkable superiority, outperforming 23 established DA methods by a significant margin. Furthermore, our in-depth analyses shed light on the decision-making processes, revealing insights into the adaptability and efficiency of GAN-DA. This approach not only addresses the limitations of existing DA methodologies but also sets a new benchmark in the realm of domain adaptation, offering broad implications for future research and applications in this field
Acoustic microstreaming and shear stress produced by the interaction of an oscillating gas bubble with a viscoelastic particle
No full textInternational audienceAn analytical theory is developed that describes acoustic microstreaming produced by the interaction of an oscillating gas bubble with a viscoelastic particle. The bubble is assumed to undergo axisymmetric oscillation modes, which can include radial oscillation, translation and shape modes. The oscillations of the particle are excited by the oscillations of the bubble. No restrictions are imposed on the ratio of the bubble and the particle radii to the viscous penetration depth and the separation distance, as well as on the ratio of the viscous penetration depth to the separation distance. Capabilities of the developed theory are illustrated by computational examples. The shear stress produced by the acoustic microstreaming on the particle surface is calculated. It is shown that this stress is much higher than the stress predicted by Nyborg's formula [J. Acoust. Soc. Am. 30, 329-339 (1958)], which is commonly used to evaluate the time-averaged shear stress produced by a bubble on a rigid wall.</div