Archive ouverte de Centrale Lyon
Not a member yet
32420 research outputs found
Sort by
Leveraging the integration of perovskite BaTiO<sub>3</sub> on ferroelectric fluorite HfO<sub>2</sub> to enhance energy storage cyclability and efficiency
No full textInternational audiencePerovskite relaxor ferroelectrics and antiferroelectrics have been the workhorse materials for energy storage applications. Recently, there has been growing interest in ferroelectric HfO2 , which is highly integrable and scalable down to the ultrathin limit. However, wake-up and fatigue phenomena in ferroelectric HfO2 are serious limitations. Interface engineering using simple AxOy binary oxides as capping layers is a promising strategy to mitigate these effects. Instead of using AxOy, we demonstrate that the integration of perovskite ferroelectric BaTiO3 as capping layer of fluorite ferroelectric Hf0.5Zr0.5O2 allows to obtain a highly cyclable and stable energy storage device of enhanced efficiency on Si(001). We study a set of samples combining epitaxial Hf0.5Zr0.5O2 with polycrystalline BaTiO3 of various thicknesses. The ferroelectricity of Hf0.5Zr0.5O2 is preserved and the efficiency is enhanced without compromising its high breakdown voltage. The device performance is optimal in terms of energy storage capacity, efficiency and breakdown field for a BaTiO3 thickness of 10 nm. We discuss the improved efficiency in terms of the role of depolarizing electric fields and the improved stability in terms of the smoother device band diagram owing to the BaTiO3 presence. These particular characteristics are attained thanks to the distinct electrical properties of BaTiO3 compared with previously investigated simpler AxOy capping layers. The investigations presented here can help to develop new strategies to enhance the energy storage efficiency in hafnia-based devices fully compatible with industrial processes
Patch-based Representation and Learning for Efficient Deformation Modeling
No full textInternational audienc
Target controllability for a minimum time problem in a trait-structured chemostat model
No full textIn this paper, we consider a minimum time control problem governed by a trait-structured chemostat model including mutation and one limiting substrate. Our first main result proves the well-posedness of the control-to-state mapping. We subsequently analyze the class of {\it{auxostat-type controls}}, feedback laws designed to regulate substrate concentration, and prove that the corresponding solutions converge to a stationary state of the system. These convergence results are used to show the reachability of a target set corresponding to the selection of a population with a low weighted averaged half-saturation constant. Finally, we show the existence of an optimal control for the minimum time problem associated with reaching the target set. These theoretical findings are completed by numerical simulations
CIP-Net: Continual Interpretable Prototype-based Network
No full textInternational audienceContinual learning constrains models to learn new tasks over time without forgetting what they have already learned. A key challenge in this setting is catastrophic forgetting, where learning new information causes the model to lose its performance on previous tasks. Recently, explainable AI has been proposed as a promising way to better understand and reduce forgetting. In particular, self-explainable models are useful because they generate explanations during prediction, which can help preserve knowledge. However, most existing explainable approaches use post-hoc explanations or require additional memory for each new task, resulting in limited scalability. In this work, we introduce CIP-Net, an exemplar-free self-explainable prototype-based model designed for continual learning. CIP-Net avoids storing past examples and maintains a simple architecture, while still providing useful explanations and strong performance. We demonstrate that CIP-Net achieves state-of-the-art performances compared to previous exemplar-free and self-explainable methods in both task-and class-incremental settings, while bearing significantly lower memory-related overhead. This makes it a practical and interpretable solution for continual learning
Influence of the Reynolds number on non-Newtonian flow in thin porous media
No full textWe study the effect of the Reynolds number on the flow of a generalized Newtonian fluid through a thin porous medium in R 3 . This medium is a domain of thickness ε ≪ 1, which is perforated by periodically distributed solid cylinders of size ε. We consider the nonlinear stationary Navier-Stokes system with viscosity following the Carreau law. Using the tools of homogenisation theory and assuming that the Reynolds number scales as ε -γ , where γ is a real constant, we prove the existence of a critical Reynolds number of order 1/ε, in the sense that the inertial term in the Navier-Stokes system has no influence in the limit if the Reynolds number is of order smaller than or equal to 1/ε (i.e., γ = 1). In this case, we derive linear or nonlinear Darcy's laws connecting velocity to pressure gradient. Conversely, we expect a contribution from the inertial term in the homogenised problem if the Reynolds number is greater than 1/ε. Finally, we propose a numerical method to solve nonlinear Darcy's law describing effective flow in the critical case and demonstrate its practical applicability by applying it to several examples
Construction few-shot et fusion de graphes de connaissances temporels dynamiques atomiques à partir de textes
No full textNational audienceAvec la croissance rapide des données, extraire des connaissances de textes non structurés est devenu crucial pour l'analyse en temps réel, l'inférence temporelle et les mémoires dynamiques des agents. Pourtant, les graphes de connaissances traditionnels négligent souvent l'évolution des données, limitant leur adaptabilité. De plus, les approches récentes utilisant des modèles de langage en zero-shot ou few-shot, sans fine-tuning ni ontologies spécialisées, restent sujettes à l'instabilité et à une couverture incomplète des faits. Pour surmonter ces limites, nous proposons ATOM , une méthode few-shot et scalable pour construire et mettre à jour en continu des graphes de connaissances temporels à partir de textes non structurés. Notre approche segmente les documents en « faits atomiques » afin d'améliorer l'exhaustivité et la stabilité, génère des graphes temporels atomiques via une modélisation temporelle duale distinguant observation et validité, puis fusionne ces graphes en parallèle. Les résultats empiriques montrent des gains d'environ 18% en exhaustivité, 17% en stabilité, et plus de 90% de réduction de latence par rapport aux méthodes de référence
Évaluation de la segmentation des cavités nasales à l'aide d'une métrique topologique Betti-1
No full textInternational audienc
Experimental Demonstration of an IGCT/SiC PiN Diode Switching Cell Operating Without Any Turn-On Snubber
No full textInternational audienceIGCTs are attractive for high power, low-frequency converters such as HVdc MMCs because of their lower losses compared to IGBTs. However, a drawback of IGCTs is that they typically require an additional and bulky turn-on snubber circuit to limit their turn-on dI/dt. This limitation in current dynamics prevents destructive recovery from occurring in the diode of an IGCT/diode switching cell. Previous studies have shown the possibility of removing the snubber by setting strict limits to the operating range of the IGCT/Si-diode pair, which in practice strongly restrain the application of a snubberless IGCT submodule or switching cell.In this article, high-voltage, high-current, SiC PiN diodes are used successfully with IGCTs in a snubberless configuration up to 4 kV/500 A, and a comparison is made with IGCTs operating with Si PiN diodes. The IGCT turn-on behavior is analyzed, through the evaluation of the IGCT equivalent resistance during turn-on. The impact of very short reverse recovery (RR) is also investigated, with the observation and analysis of the shared voltage between the IGCT and the diode and its impact. It is found that SiC PiN diodes can operate with IGCTs without the limitations encountered with Si diodes and without requiring a turn-on snubber
Multi-serotype nested immuno-epidemiological model for dengue hemorrhagic fever involving backward bifurcation and serotype invasion
No full textInternational audienceReinfection with the same dengue serotype is generally benign, as individuals develop protective immunity. On the other hand, in the case of reinfection with a different serotype, pre-existing antibodies can increase the risk of developing Dengue Hemorrhagic Fever (DHF), by inducing Antibody-Dependent Enhancement (ADE). To model this dynamic, we introduce a multi-scale immuno-epidemiological system. The immunological part is described by a system of ODEs representing the interaction between two antibodies (from previous and current infection) and the virus. The epidemiological part is represented by an infection-age structured SIRS system (for both the {\color{red} primary and secondary} infections) and a recovery-age structured equation (for the first infection). A detailed mathematical analysis of the equilibrium points of the multi-scale reinfection model, including disease-free, mono-endemic and bi-endemic states, is performed. We establish necessary and sufficient conditions for the existence of backward bifurcations and derive an expression for the invasion reproduction number, which shows that the second serotype can invade the population after a mono-endemic first serotype. We also investigate the dependence of the basic and invasion reproduction numbers on the immunological parameters of the first and second infections. This gives us a better understanding of the relationship between DHF and ADE during secondary infection
Backstepping for partial differential equations: A survey
No full textInternational audienceSystems modeled by partial differential equations (PDEs) are at least as ubiquitous as those by nature finite-dimensional and modeled by ordinary differential equations (ODEs). And yet, systematic and readily usable methodologies, for such a significant portion of real systems, have been historically scarce. Around the year 2000, the backstepping approach to PDE control began to offer not only a less abstract alternative to PDE control techniques replicating optimal and spectrum assignment techniques of the 1960s, but also enabled the methodologies of adaptive and nonlinear control, matured in the 1980s and 1990s, to be extended from ODEs to PDEs, allowing feedback synthesis for systems that are uncertain, nonlinear, and infinite-dimensional. The PDE backstepping literature has since grown to hundreds of papers and nearly a dozen books. This survey aims to facilitate the entry into this thriving area of overwhelming size and topical diversity. Designs of controllers and observers, for parabolic, hyperbolic, and other classes of PDEs, in one or more dimensions, with nonlinear, adaptive, sampled-data, and event-triggered extensions, are covered in the survey. The lifeblood of control are technology and physics. The survey places a particular emphasis on applications that have motivated the development of the theory and which have benefited from the theory and designs: flows, flexible structures, materials, thermal and chemically reacting dynamics, energy (from oil drilling to batteries and magnetic confinement fusion), and vehicles