Portail HAL Ensta
Not a member yet
11080 research outputs found
Sort by
Contrôle Optimal et Robuste pour le Guidage de Véhicules Autonomes
No full textThe guidance of a reusable launcher is a control problem that requires both precision and robustness: one must compute a trajectory and a control such that the system reaches the landing zone, without crashing into it or exploding mid-flight, all while using as little fuel as possible. Optimal control methods based on Pontryagin's Maximum Principle can compute an optimal trajectory with great precision, but uncertainties, the discrepancies between estimated values of the initial state and parameters and actual values, cause the actual trajectory to deviate, which can be dangerous. In parallel, set-based methods and notably validated simulation can enclose all trajectories of a system with uncertainties. This thesis combines those two approaches to enclose sets of optimal trajectories of a problem with uncertainties to guarantee the robustness of the guidance of autonomous vehicles. We start by defining sets of optimal trajectories for systems with uncertainties, first for mathematically perfect trajectories, then for the trajectory of a vehicle subject to estimation errors that can use, or not use, sensor information to compute a new trajectory online. Pontryagin's principle characterizes those sets as solutions of a boundary value problem with dynamics subject to uncertainties. We develop algorithms that enclose all solutions of these boundary value problem using validated simulation, interval arithmetic and contractor theory. However, validated simulation with intervals is subject to significant over-approximation that limits our methods. To remedy that we replace intervals by constrained symbolic zonotopes. We use those zonotopes to simulate hybrid systems, enclose the solutions of boundary value problems and build an inner-approximation to complement the classical outer-approximation. Finally, we combine all our methods to compute sets of trajectories for aerospace systems and use those sets to assess the robustness of a control.Le guidage d'un lanceur réutilisable est un problème de contrôle qui nécessite à la fois précision et robustesse : il faut calculer une trajectoire et un contrôle, de sorte que le lanceur atteigne la piste d'atterrissage, sans s'écraser ni exploser en vol, le tout en utilisant le moins de carburant possible. Les méthodes de Contrôle Optimal issu du Principe de Pontryagin calculent une trajectoire optimale avec grande précision, mais les incertitudes, soit les erreurs entre les estimations de l'état initial et des paramètres et leurs valeurs réelles, causent une déviation potentiellement dangereuse de la trajectoire réelle. En parallèle, les méthodes ensemblistes et notamment la simulation validée peuvent encadrer toutes les trajectoires possibles d'un système dynamique avec des incertitudes bornées. Cette thèse combine ces deux approches pour encadrer des ensembles de trajectoires optimales de systèmes avec incertitudes afin de garantir la robustesse du guidage d'un véhicule autonome. Nous commençons par définir des ensembles de trajectoires optimales pour des systèmes avec incertitudes, d'abord pour les trajectoires mathématiquement parfaites, puis pour les trajectoires d'un véhicule sujet à des erreurs d'estimation, mais qui utiliserait, ou non, les données des capteurs pour recalculer sa trajectoire en cours de route. Le principe de Pontryagin caractérise ces ensembles comme solutions de problèmes aux deux bouts avec des dynamiques avec incertitudes. Nous développons alors des algorithmes qui encadrent toutes les solutions de ces problèmes aux deux bouts en utilisant la simulation validée, l'arithmétique des intervalles et la théorie des contracteurs. Cependant, la simulation avec des intervalles occasionne une forte sur-approximation qui limite nos méthodes. Pour y remédier, nous remplaçons les intervalles par des zonotopes symboliques contraints. Nous utilisons notamment ces zonotopes pour simuler des systèmes hybrides, encadrer des solutions de problèmes aux deux bouts et construire des sous-approximations en complément de la sur-approximation classique. Enfin, nous combinons tout ceci pour calculer des ensembles de trajectoires de systèmes aérospatiaux et les utilisons pour évaluer la robustesse du contrôle
A multi-physics, multi-scale and finite strain crystal plasticity-based model for pseudoelastic NiTi shape memory alloy
No full textInternational audienc
Measurements of pressure during the forced water entry of a cone into pure and aerated water
No full textInternational audienc
Porous Composite Nanomaterials based on MOFs for green Hydrogen Production by Photocatalysis,
No full textInternational audienc
Porous Composite Nanomaterials based on MOFs for green Hydrogen Production by Photocatalysis,
No full textInternational audienc
Pushing the Frontier in the Design of Laser-Based Electron Accelerators with Groundbreaking Mesh-Refined Particle-In-Cell Simulations on Exascale-Class Supercomputers
No full textInternational audienc
Recueil des actes de la 12ème conférence sur les vibrations induites par un écoulement.
No full textParis-Saclay 5-8 July 2022International audienceFlow-Induced Vibration remains a challenge for design engineers and operators acrossmultiple industries, including vehicle manufacturing and power production, among manyothers. Despite decades of research and development, there is still much to be learned inthis domain. Specific areas warrant special attentions, such as fluid-elastic instability oftube bundles, multiphase flows and their effects on flexible structures, vortex shedding intheir many manifestations, and the intricacies of axial flow instabilities. Even seeminglysimple scenarios, like a single cylinder in crossflow, continue to intrigue researchers,leading to the publication of new scientific articles
Modelling of the compressible hyperelastic behavior of polyurethane foams: analysis of some models and of the influence of the identification strategy
No full textInternational audiencePolyurethane foams are widely used in the automotive industry. In jounce bumpers, this material is subjected to complex multiaxial loadings. Numerical simulations used to design such parts require constitutive models that accurately and reliably describe the mechanical behavior of the material over a wide range of loading conditions. This paper deals with the modelling of the highly compressible hyperelastic behavior of polyurethane foams. It investigates the hyperfoam model, widely used today in the industry. In particular, the influence of the identification strategy, is examined. The identification of models parameters uses a database composed of uniaxial compression, uniaxial tension, simple shear and confined compression tests. The capacity of the model to predict the response to these loadings according to the experimental data used to identify its parameters is investigated. Finally, simulations of a polyurethane foam jounce bumper under global axial compression are performed. The results are compared to the experimental load-deflection curve and to the deformed shape of the bumper obtained by X-Ray tomography
Effect of Active Plastic Fine Fraction on Undrained Behavior of Binary Granular Mixtures
No full textInternational audienceThe mechanical behavior of binary granular mixtures strongly depends on their initial packing density, stress level, fine content, particle size distribution, mineralogy, and shape. This research aims to investigate the effect on the mechanical behavior of fine-sand mixtures of fine particle fraction through various features: grain size distribution, fine particle size, and plasticity. The concept of equivalent intergranular void ratio is proposed for this analysis. It is correlated to the micromechanical activation of fines within the sand matrix.Monotonic consolidated undrained triaxial tests are carried out for mixtures of coarse particles (sand) and fine particles (silt or clay), in the sand dominant behavior, having various shapes and grain-size distributions. Loose, medium, and dense mixtures are tested using different fine contents and confining pressures. The undrained response is strongly affected by particle interactions, depending on the packing density, particle size, and plasticity.The active fine fraction captures the active contribution of fine particles in the sand skeleton structure. It influences the equivalent intergranular void ratio estimated in these experiments and associated to the steady state of mixtures. The reliability of equivalent state theory and an original formula proposed to estimate the active fine fraction is demonstrated in the case of fine-sand mixtures having plastic fine particles and confirmedfor non-plastic fines
