Portail HAL Ensta
Not a member yet
11080 research outputs found
Sort by
Performance amelioration of simultaneous up- and down-mixing contingent on a parallel SOA-MZIs design
No full textInternational audienceThe primary focus of this research is the creation of a first-ever experimental investigation for simultaneous up-down mixing based on three parallel semiconductor optical amplifier Mach-Zehnder interferometers (SOA-MZIs). The three mixed signals from the SOA-MZI terminals are coupled together to create a simultaneous up or down-mixed signal using an optical coupler (OC). The identified mixing function makes use of the consequences of cross-gain modulation (XGM) as well as cross-phase modulation (XPM) that have been shown at the SOA-MZI outlets. Furthermore, the performance characterization of the photonic parallel system is examined and assessed in the optical and electrical sectors. The outcomes of our experiment closely resemble those of the simulation. Several measurable factors, including power scaling measurements, optical conversion gain (OCG), power stability, beam quality factor, electrical conversion gain (ECG), and error vector magnitude (EVM), are employed to evaluate the photonic up and down-mixed signals. This three-parallel SOA-MZIs architecture has a wide range of applications involving radio over fiber and optical telecommunications. A 256-QAM (Quadrature Amplitude Modulation) up- and down-mixed signals can have a top bit rate of 64 Gbit/s with excellent EVM values. Additionally, the maximal frequency range of the concurrent up and down-mixed signal is 80 GHz. The ECG also displays significant values for up and down conversion. In the optical domain, the OCG has the highest values of 15 dB for the up-mixed signal and 12 dB for the down-mixed one, with a data input power of −10 dBm. Finally, the simultaneous mixing system tackles an outstanding beam quality factor with M2 measurement of less than 1.3 (M2)
Un algorithme branch-and-bound pour résoudre exactement des problèmes d'optimisation parcimonieuse structurée
No full textInternational audienceLes problèmes d'ajustement de modèles de faible cardinalité ont trouvé de nombreuses applications en statistique, en optimisation de portefeuille et en traitement du signal.Parmi ces problèmes, nous nous intéressons à la résolution d'un problème structuré, dans lequel la pénalité de faible cardinalité l0 portesur des groupes de composantes et non sur les composantes individuelles, et l'ajustement au modèle se fait par moindres carrés.Ce problème, plus général que son équivalent non structuré, ne possède pas à notre connaissance de méthode dédiéepour trouver son optimum global de manière garantie.Nous proposons dans cette communication d'utiliser une méthode de branch-and-bound,et de calculer les bornes inférieures de chaque nœud via une relaxation l1 du problème originelpermettant à la fois de gérer des cas de groupes se chevauchant et de réutiliserdes algorithmes et techniques d'accélérations spécifiques aux problèmes l1 standard.Nous intégrons cette approche dans une extension du solveur open-source Mimosa pour résoudre exactement ce problème structuré
In-Situ Monitoring Of The Ultrastructure And Mechanical Properties Of Flax Cell Walls During Controlled Heat Treatment
No full textInternational audiencePlant fibres are increasingly used as reinforcements, especially in thermoplastic composites. Understanding the impact of temperature on the properties of these fibres is an important issue for the manufacturing of high-performance materials with minimal defects. In this work, the structural evolution and mechanical behaviour of flax fibre cell walls were dynamically monitored by temperature-controlled X-ray diffraction and nanoindentation from 25 to 230°C; detailed biochemical analysis was also conducted on fibre samples after each heating step. With increasing temperature up to 230°C, a drop in the local mechanical performance of the flax cell walls was measured. This was associated with a decrease in the packing of the cellulose crystal lattice (increase in d-spacing d200), as well as significant mass losses measured by TGA and changes in the biochemical composition, i.e. non-cellulosic polysaccharides (NCPs) of the middle lamellae but also of the cell walls. This work, which proposes for the first time an in-situ investigation of the dynamic temperature evolution of the flax cell wall properties, evidences the reversible behaviour of their crystalline structure (i.e. cellulose) and local mechanical properties after cooling to room temperature, even after exposure to high temperatures
Nonlinear model order reduction of resonant piezoelectric micro-actuators: An invariant manifold approach
No full textInternational audienceThis paper presents a novel derivation of the direct parametrisation method for invariant manifolds able to build simulation-free reduced-order models for nonlinear piezoelectric structures, with a particular emphasis on applications to Micro-Electro-Mechanical-Systems. The constitutive model adopted accounts for the hysteretic and electrostrictive response of the piezoelectric material by resorting to the Landau-Devonshire theory of ferroelectrics. Results are validated with full-order simulations operated with a harmonic balance finite element method to highlight the reliability of the proposed reduction procedure. Numerical results show a remarkable gain in terms of computing time as a result of the dimensionality reduction process over low dimensional invariant sets. Results are also compared with experimental data to highlight the remarkable benefits of the proposed model order reduction technique
Goal Space Abstraction in Hierarchical Reinforcement Learning via Reachability Analysis
No full textInternational audienceOpen-ended learning benefits immensely from the use of symbolic methods for goal representation as they offer ways to structure knowledge for efficient and transferable learning. However, the existing Hierarchical Reinforcement Learning (HRL) approaches relying on symbolic reasoning are often limited as they require a manual goal representation. The challenge in autonomously discovering a symbolic goal representation is that it must preserve critical information, such as the environment dynamics. In this work, we propose a developmental mechanism for subgoal discovery via an emergent representation that abstracts (i.e., groups together) sets of environment states that have similar roles in the task. We create a HRL algorithm that gradually learns this representation along with the policies and evaluate it on navigation tasks to show the learned representation is interpretable and results in data efficiency
Decomposition Methods for Monotone Two-Time-Scale Stochastic Optimization Problems
No full textInternational audienceIt is common that strategic investment decisions are made at a slow time-scale, whereasoperational decisions are made at a fast time-scale. Hence, the total number of decisionstages may be huge. In this paper, we consider multistage stochastic optimization problemswith two time-scales, and we propose a time block decomposition scheme to addressthem numerically. More precisely, i) we write recursive Bellman-like equations at the slowtime-scale and ii), under a suitable monotonicity assumption, we propose computable upperand lower bounds — relying respectively on primal and dual decomposition — forthe corresponding slow time-scale Bellman functions. With these functions, we are ableto design policies. We assess the methods tractability and validate their efficiency bysolving a battery management problem where the fast time-scale operational decisionshave an impact on the storage current capacity, hence on the strategic decisions to renewthe battery at the slow time-scale
Fireball - Fourth Industrial REvolution: Blockchain & Artificial inteLLigence
No full textInternational audienc
An analytical approach of design for recycling of laminate structures by the use of magnetic pulse disassembling
No full textInternational audienceComposite materials in association with metal sheets or ceramic coatings have found a variety of applications, especially as load-bearing components in composite structures. Their disassembly for recycling or maintenance has therefore become more complex. The magnetic pulse technique, usually dedicated to dynamic forming and welding processes, can be used to separate composite materials of a laminate structure of their recyclable adjacent materials. This technology profits from Lorentz forces appearing in electrically conductive materials placed in the vicinity of conductors in which high-intensity electrical discharges are performed. These body forces act locally during the fast energy discharge, from which a stress wave propagates in the structure. The propagation of this stress wave can be used to generate interfacial tensile stress in laminate structures or assemblies. Especially, this technique can be used for evaluating the dynamic bond strength of an interface. This works aims at determining the required conditions for disassembling a laminate structure using a magnetic pulse imposed on one side of the assembly, without significantly damaging the debonded layers. Ideally, the generated interfacial tensile stress should be greater than the interfacial bond strength. In this paper, a one-dimensional analytical analysis is performed in linear elastodynamics on a tri-layered laminate structure, infinite in transverse directions, based on the method of characteristics. An optimisation problem is defined, maximising the interfacial stress between the first two layers, whose solution requires to study various configurations of the assembly involving different sets of characteristics in the laminate, associated with different areas of the domain of feasibility spanned by the unknown vector. Some assumptions and introduced simplifications of the optimisation problem allows to follow a simple two-stage solution procedure, first with respect to the thickness of the first layer for a given stress pulse, then with respect to material impedances of all layers. Several configurations of the assembly are shown to create a sufficiently large interfacial tensile stress to reach the crack initiation and propagation, and a maximum tensile interface stress of twice the maximum applied pressure is obtained in the asymptotic limit
