Portail HAL Ensta
Not a member yet
11080 research outputs found
Sort by
Finite Element Modeling and Simulation of a Submerged Wave Energy Converter System for Application to Oceanic Islands in Tropical Atlantic
No full textInternational audienceThe development of efficient and sustainable marine energy converter systems is a great challenge, especially in remote areas such as oceanic islands. This work proposes a numerical modeling methodology to assess the mechanical behavior of a wave energy converter (WEC) to be applied outside Fernando de Noronha Island (Pernambuco, Brazil). First, oceanographic data collected in situ were analyzed to determine different sea state scenarios in the region. The Airy theory and second-order Stokes’ theory were used to obtain the velocity profiles for the maximum and operational swells. These profiles were then implemented in a flow model developed in COMSOL Multiphysics software (Burlington, MA, USA) to calculate the wave distributions of pressure on the WEC structure. Finally, wave pressure distributions obtained from simulations were implemented in a static analysis of the system by the finite element method using SolidWorks (France). The results highlighted the most critical system inclination and the parts of the WEC structure more likely to be damaged under extreme swell conditions. The 0° inclination was the most critical situation, leading to the exceeding of the elastic limits of some parts of the WEC structure. The methodology developed in this work showed to be efficient to study and propose project improvement for the strength of the WEC system
The Half-Space Matching method for elastodynamic scattering problems in unbounded domains
No full textInternational audienceIn this paper, the Half-Space Matching (HSM) method, first introduced for scalar problems, is extended to elastodynamics, to solve time-harmonic 2D scattering problems, in locally perturbed infinite anisotropic homogeneous media. The HSM formulation couples a variational formulation around the perturbations with Fourier integral representations of the outgoing solution in four overlapping half-spaces. These integral representations involve outgoing plane waves, selected according to their group velocity, and evanescent waves. Numerically, the HSM method consists in a finite element discretization of the HSM formulation, together with an approximation of the Fourier integrals. Numerical results, validating the method, are presented for different materials, isotropic and anisotropic. Comparisons with the Perfectly Matched Layers (PML) method are performed for several anisotropic materials. These results highlight the robustness of the HSM method compared to the sensitivity of the PML method with respect to its parameters
Detection of AIS Messages Falsifications and Spoofing by Checking Messages Compliance with TDMA Protocol
No full textInternational audienceAutomatic identification system is an navigation aid system that allows vessels to exange automatically positions, identity and other information. It improves the safety and of the maritime traffic and help to monitor it. However, this system can be manipulated to send falsified information and mask illicit activities. While previous research has proposed strategies to detect these threats, none of them suggest a solution that considers the time-division multiple access (TDMA) communication protocol. In this work, the compliance of the sent messages with this protocol, specified by the system's standard, is checked for every ship to detect message falsifications. Furthermore, because the ships velocity affects TDMA protocol, a strategy based on the use of a Kalman filter is applied to track every ship and assesses the consistency of their velocity data sent.The proposed strategy was validated on real data, is computationally cheap and can be run in real time. Source codes of the method are open-source to foster research activities from both industry and academia in this field.</div
Guiding Pretraining in Reinforcement Learning with Large Language Models
No full textInternational audienceReinforcement learning algorithms typically struggle in the absence of a dense, well-shaped reward function. Intrinsically motivated exploration methods address this limitation by rewarding agents for visiting novel states or transitions, but these methods offer limited benefits in large environments where most discovered novelty is irrelevant for downstream tasks. We describe a method that uses background knowledge from text corpora to shape exploration. This method, called ELLM (Exploring with LLMs) rewards an agent for achieving goals suggested by a language model prompted with a description of the agent's current state. By leveraging large-scale language model pretraining, ELLM guides agents toward human-meaningful and plausibly useful behaviors without requiring a human in the loop. We evaluate ELLM in the Crafter game environment and the Housekeep robotic simulator, showing that ELLM-trained agents have better coverage of common-sense behaviors during pretraining and usually match or improve performance on a range of downstream tasks. Code available at https://github.com/yuqingd/ellm
The European Union as an actor of defence industrial policy
No full textInternational audienceSince 2016 and even more after the Russian invasion of Ukraine in 2022, the European has geared up its toolbox in order to strengthen its defence industrial capabilities and capacities. However, the EU is not a legacy actor of armament, since member states had long been prevented it from interfering in those matters.Nevertheless, the lack of domestic funding and the stalemates of the creation of a European defence market in an intergovernmental approach have provided a window of opportunity for EU institutions to jump in. The war in Ukraine has generated a sense of urgency to improve the means and effectiveness of a European approach in this domain.Indeed, being an “enlightened amateur” is no longer possible for the EU when stakes reach a higher scale. Beyond the relevance of different tools created over the past decade, what can be the coherence of this toolbox? How does this issue question the notion of defence industrial policy today? What can be at stake in a European perspective?This contribution aims to analyse the grounds for defence industrial policy in arms-producing countries in the 21st century and to look at the specific European market by assessing how states and the EU are trying to build a "two-stage rocket" answering the challenges of European international security
Diffraction grating with varying slit width: Quasi-periodic homogenization and its numerical implementation
No full textInternational audienc
Semi-Analytical Load Models Accounting for the Tilt and Motion of a Cylinder Impacted by a Plunging Breaking Wave
No full textInternational audienceAbstract This work is part of the DIMPACT project that addresses breaking-wave-induced loads on floating wind turbines. In this context, the hydrodynamic impact of a strong plunging breaking wave on a cylinder is studied using semi-analytical models, with a focus on the loads sensitivity to the cylinder tilt and wave-cylinder relative motion. Two semi-analytical models are applied in a strip theory approach under the Froude-Krylov assumption. A numerical wave is generated using a Fully Nonlinear Potential Flow (FNPF) solver, whose free surface geometry and kinematics feed the semi-analytical models. Semi-analytical slamming loads are compared with force measurements collected in a wave flume for different pitch angles and surge speeds. The temporal variation of the force as well as the maximum force acting on the cylinder are well estimated by the semi-analytical models. A new engineering formula is finally developed, accounting for the inclination and surge motion of the cylinder. The impact loads computed with this formula also show good agreement with the experimental tests
Combined techniques and relevant image processing for quantitative statistical characterization of inclusions in elastomers
No full textInternational audienceABSTRACT The properties of elastomeric materials are strongly influenced by the inclusions resulting from the ingredients and the elaboration process. A methodology is proposed to differentiate the inclusions harmful for fatigue (larger than a few micrometers) in elastomers according to their chemical nature, and to characterize them quantitatively with sufficient statistics. Three techniques are used and compared: digital optical microscopy (OM), scanning electron microscopy (SEM) associated with energy dispersive X-ray spectroscopy, and X-ray micro-computed tomography (μ-CT). Six materials are used to challenge the methodology. In addition to the usual metal oxides and carbon black agglomerates, three atypical types of inclusions are highlighted, generating specific detection difficulties. A relevant image analysis procedure is developed to automatically detect the inclusions from the acquired images, more objectively and accurately than with the classical thresholding methods. The morphology and the spatial distribution of the different inclusions populations are then determined. μ-CT is the most comprehensive and accurate method for classification and statistical characterization of inclusions. Furthermore, relevant data on the size distribution of inclusions can be obtained using backscattered electrons (SEM-BSE) or digital OM. SEM-BSE provides more accurate results than digital OM
Inner and outer characterization of the projection of polynomial equations using symmetries, quotients and intervals
No full textInternational audienceIn this paper, we propose a new approach to compute the projection of a set defined by polynomial equations. It assumes that the polynomial equations have some nice symmetries and that a solution of the projection problem is already available in the case where the variables along which we project are all positive. A new interval-based algorithm which combines symmetry operators and set quotient is proposed. Symmetries are used to move from one part of the space to another. The set quotient is needed to avoid redundant symmetries. The projection procedure yields an inner and an outer approximations of the projected set. Two applications are considered. The first one corresponds to the characterization of the space occupied by a rotating polygon, and the second one deals with the estimation of the speed of a moving object observed by several robots with uncertain orientations
Ultra-wideband 3D-printed vivaldi array for in-band full-duplex applications with a grating lobes reduction approach
No full textInternational audienceThis work validates the ability for in-band full-duplex systems to operate over ultra-wide bandwidths. As a proof of concept an ultra-wideband in-band full-duplex antenna array was designed. The array is formed of two wideband microstrip-slotline baluns and four Vivaldi antennas in a 3D-printed single-piece structure. The metal thickness of the 3D-printed Vivaldi antennas makes them sturdier than PCB antennas and more immune to fracturing or breaking, however, they require a coax-to-slot transition as a feeding mechanism. Such transitions are not well documented in the literature, specifically, the effect of metal thickness on the impedance of the 3D-printed slotline. Therefore, impedance curves for the 3D slot are provided as a function of its width and metal thickness. Then, a prototype was realized and measured, achieving about 40 to 80 dB of self-interference cancellation in the frequency range from 4.4 GHz to 28.7 GHz, and an average gain of 9.8 dBi with a directive end-fire radiation pattern. The only drawback of this first prototype is the presence of grating lobes in its radiation pattern due to the separation distance between the opposite antennas which exceeds a half-wavelength. Thus, to mitigate those grating lobes, we propose an approach based on a partial size reduction at the bottom of the antenna. This allows bringing the antennas closer together at their bottom parts, reducing the distance between the radiating slots, which translates to grating lobes reduction especially at higher frequencies. The implementation of this approach results in keeping all sidelobes 10 dB below the main lobe level, while keeping the Figure of Merit of this 2nd ultra-wideband in-band full-duplex antenna system at a higher level compared to existing techniques