Portail HAL Ensta
Not a member yet
11080 research outputs found
Sort by
Influence of shot-peening on the self-heating behavior and fatigue properties of 300M steel
No full textInternational audienceShot peening is an established cold working process used to introduce residual compressive stresses on a surface and is extensively studied using conventional fatigue tests. However, it has not been widely studied using the self-heating method. Specifically, the heterogeneity of the dissipation field has not been estimated, with only an average approach being used. Previous investigations in the case of 300M steel demonstrated that the effect of shot peening on the high cycle fatigue properties can be either beneficial or detrimental. This study proposes to apply the self-heating method on polished 300M and to investigate the effect of mean stress and shot peening on the dissipation behavior. A modified self-heating model is proposed and calibrated for 300M steel. Combined with residual stress profiles, a method to compute and determine the shot peening effect on self-heating behavior through single point surface measurements is proposed. Application on 300M steel shows excellent results, the over-dissipation being mainly due to the sub-surface compressive residual stresses. The self-heating method has proven useful to quickly estimate fatigue properties of polished 300M steel. Based on the understanding of the self-heating curve of shot peened 300M steel, a quantification of shot-peening effect on fatigue limit is discussed.</div
Development of a New Laboratory Earthquake Setup Featuring a Paraffin oil-based Gel as Analogue Material
No full textInternational audienceIn this study, we present the development of a new experimental setup composed of an analogue fault surrounded by paraffinoil-based gel, which allows us to simulate earthquake-like events in the laboratory. The apparatus is designed to test the possibility ofmitigating earthquake-like instabilities using control theory. We present the physical properties of the paraffin oil-based gel as functionsof strain rate, strain, and temperature. Our results show a linear relation between the stress and strain up to 30% shear strain, along witha low viscosity at high strain rate. Furthermore, we engineered the frictional properties of the analogue fault using 3D-printed patchesplaced along its surface. Finally, an experimental earthquake simulation, using the setup, demonstrates a sudden slip event within the gel,propagating at a speed between cs and 1.41 cs, where cs represents the shear wave velocity of the gel, which is consistent with theoreticaland previous experimental results
Mesures robotisées : caractérisation acoustique automatisée des structures
No full textThis thesis aims at developing a tool for automated acoustic characterization using robotized measurements. Leveraging advancements in robotics, mechanics, and applied mathematics, the work focuses on employing a robotic arm equipped with sensors to estimate the sound field emitted by unknown structures. Since the accuracy of the measurements depends on the precise positioning of the sensor, discrepancies between the nominal and actual models of the robotic arm must be corrected. To address this, a comprehensive geometric calibration procedure is proposed and experimentally validated through two case studies involving three robotic arms and two types of sensors. Notably, significant improvements in accuracy are achieved by incorporating task-specific and sensor-infused features. The theoretical and numerical foundations of Sound Field Estimation (SFE), which aims at reconstructing the acoustic field radiated by an unknown source based on a limited set of measurements, are then explored. Among the reviewed methods, the Boundary Elements Method (BEM) stands out for its efficiency and flexibility despite its computational cost. Numerical simulations allow for the validation of the method's performances and the analysis of its robustness in the presence of external disturbances, highlighting the importance of robot calibration to ensure reliable results. Finally, to support the experimental implementation of our tool, a framework based on the ROS middleware is introduced to manage interfaces between the robot, sensors, and their environment. Prior to acoustic measurements, a robotized geometric characterization process is presented to determine the shape and position of the studied object. The robotized setup developed to automate acoustic measurements is then presented, with particular focus on the effects of the robot on the measurements. Under conditions where these effects are minimized, two autonomous measurement campaigns led to a successful reconstruction of the sound field radiated by an unknown loudspeaker, demonstrating the potential of the proposed tool.Cette thèse a pour but le développement d'un outil de caractérisation acoustique automatisé à l'aide de mesures robotisées. En s'appuyant sur des avancées en robotique, mécanique et mathématiques appliquées, les travaux réalisés traitent de l'utilisation d'un bras robotisé équipé de capteurs pour estimer le champ sonore émis par des structures inconnues. La précision des mesures dépendant de l'exactitude du positionnement du capteur, les écarts entre les modèles nominaux et réels du bras robotisé doivent être compensés. Pour cela, une procédure complète d'étalonnage géométrique est proposée et validée à travers deux études de cas impliquant trois bras robotisés et deux types de capteurs. En particulier, des améliorations significatives de la précision sont obtenues en intégrant des aspects spécifiques à la tâche visée et aux capteurs utilisés. Les fondements théoriques et numériques de l'estimation de champ sonore, qui vise à reconstruire les champs acoustiques rayonnées par une source inconnue à partir d'un ensemble limité de mesures, sont ensuite explorés. Parmi les méthodes examinées, la méthode des éléments de frontières se distingue par son efficacité et sa flexibilité, malgré ses exigences computationnelles. Des simulations numériques permettent la validation des performances de la méthode et l'analyse de sa robustesse en présence de perturbations externes, soulignant l'importance de l'étalonnage du robot pour garantir des résultats fiables. Finalement, afin d'assurer la mise en œuvre expérimentale de notre outil, un "framework" reposant sur le middleware ROS est introduit pour gérer des interfaces entre le robot, les capteurs et leur environnement. En amont des mesures acoustiques, un procédé de caractérisation géométrique robotisé permettant de déterminer la forme et la position de l'objet étudié est présenté. Le système de mesures acoustiques robotisées est ensuite décrit, avec une attention particulière portée aux effets du robot sur les mesures réalisées. En se plaçant dans des conditions où ces impacts sont limités, deux campagnes d'acquisition autonomes nous permettent de reconstruire avec précision le champ sonore rayonné par une enceinte inconnue, démontrant ainsi le potentiel de l'outil développé
Fissuration par champ de phase : Suivre le chemin d'équilibre de la structure
No full textInternational audienc
Méthode ellipsoïdale numérique garantie pour l’analyse de la stabilité du contrôle de formation d’un groupe de robots sous-marins
No full textIn the development of human marine activity, groups of underwater robots can automate certain tasks. Since these robots are difficult to localise because of the underwater constrains, they must move in formation to be reliable. While various theoretical controllers have been proposed to challenge these constrains, they still need to consider more complex constrains and to be tested on real systems. As for every autonomous system, the stability of the formation must be verified by a mathematical proof. However, the complexity of these nonlinear systems makes conventional Lyapunov method difficult to use. Thus, this thesis’ main objective is to develop guaranteed numerical methods, based on interval arithmetic, that can assist the stability proof. Based on ellipsoidal guaranteed propagation, a first method is designed for discrete time systems to compute an ellipsoidal domain of attraction. This method is then extended to continuous-time systems and then to synchronous hybrid systems which are more realistic modellings. In addition, the ellipsoidal propagation is extended to consider singular mappings and degenerate ellipsoids. Finally, some real world underwater formation control was achieved to illustrate the stability.Pour développer les activités marines humaines, des groupes de robots sous-marins peuvent automatiser certaines tâches. Ces robots étant difficiles à localiser en raison de contraintes sous-marines, ils doivent se déplacer en formation pour être fiables. Bien que plusieurs contrôleurs théoriques aient été proposés pour faire face à ces contraintes, ils doivent encore s’adapter à des contraintes plus complexes et être testés sur des systèmes réels. Comme pour tout système autonome, la stabilité de la formation doit être vérifiée par une preuve mathématique. Cependant, la complexité de ces systèmes non linéaires rend la méthode de Lyapunov conventionnelle difficile à utiliser. Ainsi, l’objectif principal de cette thèse est de développer des méthodes numériques garanties, basées sur l’arithmétique des intervalles, qui peuvent assister la preuve de stabilité. Basée sur la propagation garantie ellipsoïdale, une première méthode est conçue pour les systèmes à temps discret afin de calculer un domaine d’attraction ellipsoïdal. Cette méthode est ensuite étendue aux systèmes à temps continu, puis aux systèmes hybrides synchrones, qui sont des modélisations plus réalistes. En outre, la propagation ellipsoïdale est étendue pour prendre en compte les applications singulières et les ellipsoïdes dégénérées. Enfin, des test de formation en situation réelle viennent illustrer la stabilité
A Mixed audio-video SPD network for online classification of Parkinsonian speech patterns
No full textInternational audienceParkinson's disease (PD) is a neurodegenerative disease that produces progressive motor impairments. Dysarthria (speech disorders) and hypomimia (face rigidity) are two major Parkinsonism patterns observed even at the early stages of the disease. Nonetheless, the clinical diagnosis is mainly observational and dependent on the specialists' expertise. Besides, the categorization of each of these patterns is isolated, which may lead to delayed diagnosis and misplanning of treatments. This work introduces a non-invasive multimodal strategy that integrates video and audio modalities into the online characterization of speech exercises. Subjects were invited to pronounce sustained vowels while video and audio were recorded. Then, a temporal window is run along the sequence to build online covariance matrices of synchronized face landmarks position and characteristic voice frequencies. From these temporal covariance matrices are learned Riemannian descriptors that allow to discriminate between Parkinson's and control subjects. From a study with 14 subjects, the proposed approach achieved a mean accuracy of 70% in sustained vowel pronunciation. Considering online predictions, the proposed approach evidenced a consistent accuracy of 0.77 during pronunciation of close vowels.</div
Tw-class Sub-2-cycle post-compression of multi-mJ energy Ti:sapphire laser pulses in a gas-filled multi-pass cell
No full textInternational audienceWe report on the nonlinear temporal post-compression of 7 mJ sub-40 fs pulses from a commercial kHz Ti:sapphire laser down to a record 3.8 fs duration (sub-1.5 optical cycle) in a compact single-stage gas-filled multi-pass cell (MPC), with 60% overall compression efficiency
Toward Using Monostatic Antennas with Near-Field Cancellation Technique in IBFD Phased Arrays
No full textBest Paper Award !International audienceThis paper presents an S-Band monostatic 2x2 patch array, with shared radiating elements among Tx and Rx, for In-Band Full-Duplex (IBFD) applications. In this array, self-interference cancellation (SIC) is achieved by locally implementing the near field cancellation (NFC) technique at each single patch by a multi-point differential feeding, which also ensures global SIC at the array level. Moreover, to identify the major coupling paths between array ports that contribute to cancellation, we provide some coupling matrices as a new way to represent the magnitude and phase combinations of interand intra-port coupling. In addition to that, we demonstrate the variation of cancellation level as the Tx and Rx beams are steered independently
Predicting the Adhesive Layer Thickness in Hybrid Joints Involving Pre-Tensioned Bolts
No full textInternational audienceWhile most academic studies focus on the properties of cured joints, this research addresses the manufacturing process of hybrid joints in their uncured state. Hybrid joints that combine adhesive bonding with pre-tensioned bolts exhibit superior mechanical performance compared to exclusively bonded or bolted joints. However, the adhesive flow during manufacturing in hybrid joints often results in a nonuniform adhesive thickness, where obtaining an exact thickness is crucial for accurate load capacity predictions. This paper presents experiments involving three different adhesives, providing precise measurements of the adhesive layer thickness distribution, which served as a reference when evaluating and validating the subsequent numerical predictions. The numerical predictions were performed using computational fluid dynamics (CFD) to model the flow behavior of the adhesives during the bonding process and their interactions with the metal substrates. The CFD predictions of the adhesive layer thickness showed good agreement with the experimental data, with the relative differences between the average experimental and numerical thickness values ranging from 4.07% to 27.1%. The results were most accurate for the adhesive with sand particles, whose particles remained intact, ensuring that the adhesive’s rheology remained unchanged. The results highlight the importance of the rheological behavior of the adhesive in the final distribution of the adhesive layer thickness, thereby expanding the understanding of these joints
Policy Learning with a Language Bottleneck
No full textModern AI systems such as self-driving cars and game-playing agents achieve superhuman performance, but often lack human-like features such as generalization, interpretability and human inter-operability. Inspired by the rich interactions between language and decision-making in humans, we introduce Policy Learning with a Language Bottleneck (PLLB), a framework enabling AI agents to generate linguistic rules that capture the strategies underlying their most rewarding behaviors. PLLB alternates between a rule generation step guided by language models, and an update step where agents learn new policies guided by rules. In a two-player communication game, a maze solving task, and two image reconstruction tasks, we show that PLLB agents are not only able to learn more interpretable and generalizable behaviors, but can also share the learned rules with human users, enabling more effective human-AI coordination