Portail HAL Ensta
Not a member yet
11080 research outputs found
Sort by
A new adjustable localization operator method to compute local stresses and strains at notched bodies under multiaxial cyclic loading
No full textInternational audienceSimplified methods for elastoplastic calculations offer a cost-effective and time-saving solution to determine the mechanical state at the notch tip of a structure. Recently, a method using an Adjustable Localization Operator (ALO) has been developed, drawing inspiration from Eshelby's inclusion problem. This method has demonstrated its efficiency in addressing various engineering challenges owing to its robustness and inherent ability to handle multiaxial conditions. This paper aims to propose an alternative non-linear formulation of the ALO method to improve the accuracy of mechanical state predictions. To achieve this goal, Buczynski-Glinka's heuristic, which provides a non-linear plasticity correction, is expressed within the framework of the ALO method to highlight the ingredients required for a non-linear plastic correction. Subsequently, the parameters of the ALO method are adjusted based on these findings. To validate the approach, two cases are studied: a notched structure under plain strain conditions, and a more complex saddle notched structure subjected to local multiaxial proportional loadings. In both cases, the non-linear ALO method has shown a greater capability to reproduce the finite element reference results compared to the linear version of the method or to energetic methods
Influence of winding angles on hoop stress in composite pressure vessels: Finite element analysis
No full textInternational audienceVarious pressure vessels from Type I to Type IV are used to store hydrogen and compressed natural gas (CNG) at high pressures. The polymeric liner in Type IV composite pressure vessels (CPVs) does not share the load, unlike those with metallic liners (Type III); hence, the composite layers share the entire load. The winding angles play a crucial role in tailoring the pressure-bearing properties of CPVs. This study investigated the effect of fiber winding angles on the hoop stress in Type IV CPVs using finite element (FE) analysis, and the advantage of multi-angle winding was realized. Stress analysis is vital to ensure that the vessel can withstand the intended operating conditions without undergoing failure. A series of finite element analyses (FEA) was carried out using the Abaqus FE simulation software to study the behavior of an anisotropic fiber-reinforced CPV with varied winding schemes. A stress analysis was performed, and various winding schemes were compared for different cases. The hoop stress in various configurations was documented to assess and compare various winding configurations
Phase-matching-free ultrashort laser pulse characterization from transient plasma lens
No full textInternational audienceA phase-matching free ultrashort pulse retrieval based on the defocusing of a laser-induced plasma is presented. In this technique, a pump pulse ionizes a rare gas providing a plasma lens whose creation time is ultrafast. A probe pulse propagating through this gas lens experiences a switch of its divergence. The spectrum of the diverging part, isolated by a coronograph, is measured as a function of the pump-probe delay, providing a spectrogram that allows for a comprehensive characterization of the temporal properties of the probe pulse. The method, called PI-FROSt for "Plasma-Induced Frequency resolved Optical Switching", is simple, free of phase matching constraints, and can operate in both selfand cross-referenced configurations at ultra-high repetition rate in the whole transparency range of the gas. The assessment of the method demonstrates laser pulse reconstructions of high reliability in both near-infrared and ultraviolet spectral ranges
Antoine & les molécules photo-activables contre le cancer (Podcast), Episode 9 de Voyage au centre de la thèse (Institut Polytechnique de Paris)
No full texthttps://octopus.saooti.com/main/pub/podcast/55951
Near real-time water vapour monitoring with shipborne GNSS for numerical weather prediction
No full textInternational audienceMAP-IO (Marion Dufresne Atmospheric Program Indian Ocean) is a research project that aims to collect long-term atmospheric, biological and marine observations to document the under-instrumented areas of the Indian and Southern Oceans. To achieve this, the French Research Vessel (R/V) Marion-Dufresne, based in Réunion Island in the Indian Ocean, has been equipped with around twenty autonomous instruments to take measurements along its route. Among the instruments deployed on this R/V, a GNSS antenna has been in operation since autumn 2020, providing unique and continuous measurements of atmospheric water vapour in this part of the world. The data were initially transmitted daily to shore for ultra-rapid (day +1) and rapid (day +3) routine analysis. The quality of the retrieved tropospheric delays and integrated water vapour contents was highlighted in a previous study. With more reliable transmission facilities, raw GNSS data are now transmitted hourly and can be analysed with a latency of less than 30 minutes. In this study, we provide an initial assessment of the near real-time (h+20min) processing performed as part of this project. This evaluation is based on GNSS raw data collected in August 2023 during a rotation of the R/V Marion-Dufresne in the French Austral Islands (Crozet, Kerguelen, Saint-Paul and Amsterdam Islands). Processing is performed every hour over a 24-hour window, using the raw hourly data transmitted by the R/V and the cumulative real-time ephemerides and clocks provided by JPL / GDGPS. Only the last hour of each analysis is then considered. Over this period, the estimated tropospheric delays are in good agreement with the rapid routine solution, with differences of less than 1 cm RMS. The comparison with the analysis and the 6h forecast of the Météo-France's global numerical weather prediction model Arpège highlight the benefits of these shipborne GNSS measurements for numerical weather prediction. The medium-term objective of this work is to establish an operational procedure for the assimilation of these near real-time GNSS tropopheric solutions into numerical weather prediction models
Évaluation de l'intégration des technologies d'intelligence artificielle dans les activités de défense et de l'effet de la performance du système national d'innovation sur son renforcement
No full textRISC-V Embedded AI for IDS Applications
No full textNational audienceIDSs (Intrusion Detection Systems) include more and more AI (Artificial Intelligence) engines to detect several attack types. However, in order to be efficient in both learning and inference phases, such systems must include hardware coprocessors to improve AI-related computations. In this PhD thesis, we would like to explore the capabilities of RISC-V based processors in this context. RISC-V is an open-source ISA (Instruction Set Architecture) than can be easily extended. The main goal of this thesis is to propose RISC-V extensions for an IDS embedded into collaborative and heterogeneous unmanned vehicles (submarine, marine, or aerial): it must detect abnormal behaviors and must be efficient in terms of power consumption, area and runtime overheads. Furthermore, coprocessors developed in this thesis should not introduce security breaches into the system. Finally, a proof-of-concept should be developed to demonstrate the efficiency of algorithms and hardware implementations compared to software solutions
Des métaphores pour la programmation de robots
No full textInternational audienceCet article présente, au travers de métaphores, cinq exercices de programmation Scratch de robots, évalués lors d'une compétition. En premier, le robot se déplace le long d'un carré. La métaphore est celle d'un dessin sur une feuille. Les déplacements du robot se programment en paramétrant le temps de déplacement ; la métaphore du dessin fait appel à une métaphore sous-jacente "un temps qui passe est un mouvement", cette seconde métaphore reposant sur le concept de proportionnalité. Bien que cet exercice plaise aux enfants, seulement 1/3 arrivent à le ré- soudre en autonomie. Les enfants, procédant par essai-erreur, ont du mal à établir le temps nécessaire à une rotation de 90°. En second, le robot clignote en bleu en jouant la note A4, en blanc en jouant D4. Les enfants ont appris à programmer un feu tricolore et sonore, sans qu'on leur explique que lumière et son sont gérés concurrem- ment. Cet exercice est dicile, il leur manque la métaphore de l'orchestre. Le chef d'orchestre - un ordonnanceur - distribue les ordres aux diérents instrumentistes - son ou lumière. Cette métaphore révèlerait une machine notionnelle simple et observable, qui permettrait une meilleure réussite. En troisième, le robot évite un obstacle devant lui en tournant. La méta- phore est celle de la chauve-souris qui permet d'expliquer le fonctionne- ment du capteur ultrason. L'observation du comportement d'évitement réie son algorithme qui complète la métaphore et la rend opérationnelle. En quatrième, le robot joue le rôle d'un chronomètre qui ache les mi- nutes et les secondes. La métaphore est celle d'un compteur kilométrique : les chires délent et il y a propagation des retenues. Tous les enfants détectent les cas d'erreur (i.e. les secondes dépassent 60). Peu d'enfants peuvent gérer la situation sans aide, probablement car aucun algorithme de comportement ne peut être réié de l'observation d'un compteur. Des capacités de programmation avec des variables sont nécessaires. En cinquième, le robot suit une ligne noire sur fond blanc. Ce suivi fait appel à deux capteurs noir/blanc qui permette de détecter si le robot est sur la ligne noire, partiellement sorti à gauche ou à droite, complètement sorti. En CM2, cet exercice est résolu par induction, en faisant observer un programme de suivi et en complétant un programme "à trous". En conclusion, une compétition de robotique est un dispositif qui engage et motive les élèves, qui sont des facteurs de réussite. Les métaphores aident à comprendre le fonctionnement des éléments du robot et le comportement des programmes. Elles sont utiles aussi pour les séances débranchées. Lorsque l'exercice nécessite une maitrise de la programma- tion, la métaphore doit être complétée par des algorithmes
A Novel Hybrid Grid Search and Tree Parzen Estimator for Deep Learning Hyperparameters Optimization
No full textInternational audienceHyperparameter optimization plays a crucial role in maximizing the performance of Deep Learning (DL) models, particularly in the medical field. In this study, we propose a novel hybrid approach called GS-TPE, which combines Grid Search (GS) and Tree Parzen Estimator (TPE) for optimizing the hyperparameters of DL architectures in order to enhance the vigilance states classification from the EEG signals. Our experiments demonstrate that the GS-TPE approach competes with the state of the art on multiple performance metrics, leading to significantly improved classification results. The obtained accuracy with combined one-Dimensional Convolutional Neural Network and Long Short-Term Memory (1D-CNN-LSTM) and with combined Auto-Encoder and LSTM (AE-LSTM) architectures reach 93.74 and 93.53%, respectively. The proposed GS-TPE approach shows great promise for advancing the field of medical signal analysis and enhancing the accuracy of EEG-based diagnostic systems.</div
Le rôle de la bioinformatique dans la R&D vaccins de Sanofi durant la décennie 2010
No full textInternational audienceDans le sillage de la crise du Covid 19, un grand nombre d'articles ont été publiés sur le rôle de la bioinformatique dans la recherche et le développement de vaccins (R&D). L'objectif de cet article est de montrer comment la bioinformatique et la R&D sur les vaccins ont interagi ces dernières années dans le cas d'une grande entreprise biopharmaceutique : Sanofi. D'un point de vue théorique, il s'agit d'insister, dans une perspective néo-shumpétérienne, sur les trajectoires technologiques au sein d'une entreprise, en mettant au jour la structure de sa R&D afin d'identifier la transformation des processus de production d'inventions qui s'est opérée autour des technologies vaccinales. Sur le plan méthodologique, la théorie des réseaux est appliquée aux graphes générés à partir des données de citations de brevets. Les résultats que nous avons obtenus montrent que Sanofi a tiré pleinement parti de la bio-informatique dédiée au développement des vaccins à ARNm. Cependant, entre 2011 et 2020, nous n'avons pas identifié de logique de co-développement : durant cette période, la bio-informatique reste un simple input dans le processus d'innovation vaccinale