Portail HAL Ensta
Not a member yet
    11080 research outputs found

    Étude des propriétés en fatigue polycyclique des matériaux et des structures obtenus par le procédé de fabrication additive arc-fil

    No full text
    Additive manufacturing processes act as an alternative to the production of complex geometries and high added value parts. It is the case in the naval industry for which the Wire and Arc Additive Manufacturing (WAAM) process is used to fabricate hollow structures. WAAM materials usually show a heterogeneous microstructure, internal defects, a strong residual stress field and rough asbuilt surfaces. The objective of the present work is to propose a method for fatigue life prediction taking into account both the internal defects and the rough asbuilt surfaces. The material parameters of probabilistic models are determined through the use of thermometry in fatigue tests. Firstly, it is shown that the self-heating of machined samples is not affected by the presence of rare pores, allowing for the identification of the fatigue properties of a virtually defect-free WAAM material. Secondly, the use of infrared thermography allows for the detection and the tracking of fatigue cracks on rough as-built surfaces of samples, giving useful information to the modelisation of crack propagation. Finally, these tests results are confronted to the case of a hollow WAAM structures with rough as-built internal surfaces. The fatigue testing of these structures allow to set the basis of a fatigue dimensioning method.Les procédés de fabrication additive s’imposent comme des moyens de production alternatifs pour la fabrication de pièces aux géométries complexes et à forte valeur ajoutée. C’est le cas du secteur industriel naval, où est proposée l’utilisation du procédé Wire and Arc Additive Manufacturing (WAAM) pour la fabrication de pièces pouvant être creuses. Les matériaux issus de ce procédé présentent classiquement une microstructure hétérogène, des défauts internes, un champ de contraintes résiduelles et des aspérités marquées en surface. Dans ce contexte, et afin de mettre en place une méthode de dimensionnement en fatigue polycyclique d’une structure fabriquée par WAAM, les objectifs de cette thèse sont de proposer des méthodes de prévision prenant en compte, d’une part, les défauts internes inhérents au procédé, et d’autre part, l’état de surface brut de fabrication. La démarche adoptée pour répondre à ces objectifs repose sur l’utilisation de la thermométrie au cours d’essais cycliques pour l’identification de paramètres utiles à la mise en place de modèles probabilistes adaptés à chaque problématique. Il est ainsi montré, d’une part, que l’auto-échauffement (associé au caractère dissipatif) du volume moyen d’éprouvettes usinées, insensible à la présence de rares pores, permet l’identification de propriétés en fatigue d’un matériau virtuellement sain et par étude numérique du matériau avec ses défauts internes. D’autre part, le suivi par thermo-élasticimétrie (associé au couplage thermo-élastique) de surfaces brutes de fabrication permet la détection de l’amorçage de fissure et leur suivi en vue de l’identification d’une loi de propagation. Enfin ces observations sont confrontées au cas d’application d’une structure creuse aux faces internes brutes, à partir de laquelle sont posées les bases d’une démarche de dimensionnement

    Enhanced fracture toughness of adhesive joints with doping epoxy by graphene nanoplatelets

    No full text
    International audiencet is necessary to enhance the mechanical properties of adhesives to replace conventional joint methods with adhesive bonding. Epoxy in its pure state often suffers catastrophic damage due to its obvious brittleness and low fracture toughness. In this study, the double cantilever beam (DCB - Mode I) was used to characterize the fracture toughness of graphene/DGEBA-epoxy nanocomposite adhesive in bonded aluminium alloy joints and bonded composite joints. Adhesives based on an epoxy adhesive DGEBA (Bisphenol A diglycidyl ether) reinforced with two percentages (1wt.% and 2wt.%) of graphene nanoplatelets (GNP) were prepared. In this study, one shows that the fracture toughness of adhesive nanocomposites was significantly better than neat epoxy-bonded adhesives. Both types of joints contain graphene resulting in increased fracture toughness. Therefore, the maximum fracture toughness was observed until the GNP reached 1wt.%, and then it began to decrease, but it is still higher than that of the pure adhesive joint. On the other hand, this work aims to determine the influence of interfacial interactions on the behavior of enhanced bonded joints and how graphene nanoplatelets can enhance the rigidity of the interface between the substrate and the adhesive. In addition, a numerical study using ABAQUS was performed and compared with the experiments performed on DCB. For the modeling of the damage in an assembly joint, the Cohesive Zone Model (CZM) was used for the fracture behavior of the adhesive

    Experimental Validation of a Guaranteed Nonlinear Model Predictive Control

    No full text
    International audienceThis paper combines the interval analysis tools with the nonlinear model predictive control (NMPC). The NMPC strategy is formulated based on an uncertain dynamic model expressed as nonlinear ordinary differential equations (ODEs). All the dynamic parameters are identified in a guaranteed way considering the various uncertainties on the embedded sensors and the system’s design. The NMPC problem is solved at each time step using validated simulation and interval analysis methods to compute the optimal and safe control inputs over a finite prediction horizon. This approach considers several constraints which are crucial for the system’s safety and stability, namely the state and the control limits. The proposed controller consists of two steps: filtering and branching procedures enabling to find the input intervals that fulfill the state constraints and ensure the convergence to the reference set. Then, the optimization procedure allows for computing the optimal and punctual control input that must be sent to the system’s actuators for the pendulum stabilization. The validated NMPC capabilities are illustrated through several simulations under the DynIbex library and experiments using an inverted pendulum

    Sur la formation des terriens dans l'anthropocène

    No full text
    International audienceComment éduquer les terriens ? Ou, pour être un peu plus précis, comment orienter les élèves d'aujourd'hui, et en vue de demain, dans une époque quand l'habitabilité future de la terre est à risque ? Notre situation actuelle peut être caractérisée par un double mouvement par rapport à notre relation avec la terre. D'une part, il y a une vague d'anticipation pour l'avenir dans l'espace stimulée par la renaissance actuelle de l'exploration et surtout l'économie spatiale-la naissance du Space 2.0. D'autre part, il y a une vague de peur pour l'avenir de la vie sur terre associée avec la révélation de l'anthropocène-la découverte que nous sommes en train de déréguler le fonctionnement du système terre. D'un part on en voit certaines personnes qui cherchent à fuir vers Mars, d'autre part on entend les appels pour un retour vers le sol. Pour nous, les éducateurs, le défi du présent, la tâche de former les terriens, consiste-au moins en partie-à donner une orientation à nos élèves par rapport à ces deux futurs possibles

    Hybrid Turbo Equalization for Faster-than-Nyquist Underwater Communication Systems

    No full text
    International audienceFaster-Than-Nyquist (FTN) signaling is a promising technique to enhance the UnderWater Acoustic (UWA) communications rate. However, the main issue of this technique is Inter-Symbol Interference (ISI) that needs to be tackled efficiently at the receiver side. To face this issue, we propose a new Hybrid Time-Frequency Domain Equalizer (HTDFE) for FTN UWA communication systems. In contrast to the conventional HTDFE, each Frequency Domain Equalizer (FDE) at the multiple output side is followed by a joint phase estimator and feedforwardfeedback Time Domain Equalizer (TDE). Simulations over a real underwater channel show that the proposed algorithm allows a gain in performance compared to the conventional solutions

    Amarrage collaboratif automatique et sécurisé d'un robot sur une plateforme mobile : preuve par analyse par intervalles de la faisabilité d'un problème d'amarrage sous-marin

    No full text
    The increasing number of offshore facilities triggers the need for reliable autonomous robots to perform inspection and maintenance missions, while minimising operational expenses. In order to decrease the likelihood of undesired events during a mission, mathematical tools can be used to prove a priori its feasibility. In the following, new methods based on a set-membership approach are developed and presented in this regard. First, we propose a new method to analyse stability of a discrete, continuous and hybrid uncertain system. Alternatively, we present an approach based on reachability analysis for which we developed a novel constraint programming tool to implement differential constraints. Both approaches allow predicting the behaviour of a robot before its actual deployment. These tools are illustrated with realistic examples falling in the field of localisation & control, and in particular applied to underwater docking. Furthermore, the Computer-Assisted Proofs in Dynamics (CAPD) library is introduced in a robotics context via practical examples.La multiplication des installations offshore suscite un besoin de robots autonomes fiables, capable d’effectuer des missions d’inspection et de maintenance tout en minimisant les coûts opérationnels. Pour réduire le risque d’accident pendant une mission, des outils mathématiques peuvent être utilisés pour démontrer a priori son bon déroulement. Dans cette thèse, des nouvelles méthodes reposant sur une approche ensembliste sont présentées à cet effet. Premièrement, nous proposons une nouvelle méthode pour analyser la stabilité d’un système incertain discret, continue ou hybride. Ensuite, nous présentons une approche s’inspirant de l’analyse d’atteignabilité, pour laquelle nous avons développé un nouvel outil de programmation par contraintes permettant d’implémenter des contraintes différentielles. Ces deux approches permettent de prédire le comportement d’un robot avant même son déploiement. Ces outils sont illustrés par des exemples réalistes issus des domaines de la localisation et du contrôle, appliqués au problème d’amarrage sous-marin. De plus, nous présentons la librairie CAPD dans un contexte robotique grâce à des exemples pratiques

    SLURP: Side Learning Uncertainty for Regression Problems

    No full text
    International audienc

    EXplainable Neural-Symbolic Learning (X-NeSyL) methodology to fuse deep learning representations with expert knowledge graphs: The MonuMAI cultural heritage use case

    No full text
    International audienceThe latest Deep Learning (DL) models for detection and classification have achieved an unprecedented performance over classical machine learning algorithms. However, DL models are black-box methods hard to debug, interpret, and certify. DL alone cannot provide explanations that can be validated by a non technical audience such as end-users or domain experts. In contrast, symbolic AI systems that convert concepts into rules or symbols-such as knowledge graphs-are easier to explain. However, they present lower generalisation and scaling capabilities. A very important challenge is to fuse DL representations with expert knowledge. One way to address this challenge, as well as the performance-explainability trade-off is by leveraging the best of both streams without obviating domain expert knowledge. In this paper, we tackle such problem by considering the symbolic knowledge is expressed in form of a domain expert knowledge graph. We present the eXplainable Neural-symbolic learning (X-NeSyL) methodology, designed to learn both symbolic and deep representations, together with an explainability metric to assess the level of alignment of machine and human expert explanations. The ultimate objective is to fuse DL representations with expert domain knowledge during the learning process so it serves as a sound basis for explainability. In particular, X-NeSyL methodology involves the concrete use of two notions of explanation, both at inference and training time respectively: 1) EXPLANet: Expert-aligned eXplainable Part-based cLAssifier NETwork Architecture, a compositional convolutional neural network that makes use of symbolic representations, and 2) SHAP-Backprop, an explainable AI-informed training procedure that corrects and guides the DL process to align with such symbolic representations in form of knowledge graphs. We showcase X-NeSyL methodology using MonuMAI dataset for monument facade image classification, and demonstrate that with our approach, it is possible to improve explainability at the same time as performance

    0

    full texts

    11,080

    metadata records
    Updated in last 30 days.
    Portail HAL Ensta
    Access Repository Dashboard
    Do you manage Open Research Online? Become a CORE Member to access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard! 👇