HAL Arts et Métiers
Not a member yet
14127 research outputs found
Sort by
De l'apprentissage à l'expérimentation ou comment faire évoluer nos exigences
No full textInternational audienc
Comparison between modulated-pulsed and continuous wave laser powder bed fusion on an Inconel 625 alloy
No full textInternational audienceUsing pulsed or modulated laser irradiation for laser powder bed fusion (L-PBF) manufacturing can be an attractive solution for better controlling heat input, mastering geometry, and tuning microstructures of built samples. A comparison was made between a continuous wave (CW) and a modulated pulsed wave (PW) L-PBF process, considering high pulse frequencies (50 kHz), various duty cycles, and an Inconel 625 powder. In the first step, considering the precise shape of laser pulses, the peak powers were adjusted to provide similar mean powers between PW and CW and analyze the specific effect of the pulse regime at constant mean power. In the second step, single L-PBF beads were considered. The PW regime was shown to reduce the process window on (P, V) maps, increase powder-denudated widths around scan tracks, and slightly reduce the dimensions (depth, width, area) of fusion beads. In the third step, 3D samples were manufactured with PW (with duty cycles of 0.9 and 0.7 ) and CW regimes. The PW regime was shown to slightly increase the porosity rate, reduce the mean grain size by 30-40 %, and modify the crystallographic texture
A Survey on Versatile Embedded Machine Learning Hardware Acceleration
No full textInternational audienceThis survey investigates recent developments in versatile embedded ML hardware acceleration. Various architectural approaches for efficient implementation of ML algorithms on resource-constrained devices are analyzed, focusing on three key aspects: performance optimization, embedded system considerations (throughput, latency, energy efficiency) and multi-application support. Nevertheless, it does not take into account attacks and defenses of ML architectures themselves. The survey then explores different hardware acceleration strategies, from custom RISC-V instructions to specialized PE, PiM architectures and co-design approaches. Notable innovations include flexible bit-precision support, reconfigurable PE, and optimal memory management techniques for reducing weights and (hyper)-parameters movements overhead. Subsequently, these architectures are evaluated based on the aforementioned key aspects. Our analysis shows that relevant and robust embedded ML acceleration requires careful consideration of the trade-offs between computational capability, power consumption, and architecture flexibility, depending on the application
Reservoir computing with state-dependent time delay
No full textInternational audienceWe examine a new design of reservoir computing based on a linear dynamical system subject to feedback in which a delay time depends on the system's state. Despite the apparent linearity of the system under casual perusal, the system possesses a nonlinearity that can be used for time-delay reservoir computing. We find that close multiple Hopf bifurcation points lead to a rich sawtooth-shaped transient response to input signals, which can be beneficial for the computing capabilities. We benchmark the reservoir computing system's memory capacity and performance on solving delayed XOR, Iris flower classification task, and the Santa Fe time-series prediction task. We demonstrate how the nonlinearity and the memory capacity of the system can be tuned by changing the time-delay dependence
A Comparative Study of Genetic Algorithm and Particle Swarm Optimization for Hybrid Renewable Systems with Battery and Hydrogen System
No full textInternational audienceThis paper presents a comparative study of Genetic Algorithm (GA) and Particle Swarm Optimization (PSO) for optimal sizing of two hybrid renewable energy systems: Solar with Battery and Grid, and Solar with H2 System and Grid. Real-time energy consumption data from a university is used to model these systems, aiming to minimize costs while meeting energy demands and ensuring reliability. The performance of GA and PSO is compared based on solution quality, convergence speed, and computational efficiency. Results show that GA provides robust configurations, while PSO offers faster convergence. These findings support efficient and practical hybrid system design
Durabilité en eau chlorée d’un revêtement PEAA sur fonte : impact de la microstructure
No full textSaint-Gobain Pont-à-Mousson produces ductile iron steel pipes designed for drinking water networks. These pipes are coated on their internal surface with a thermoplastic copolymer, poly(ethylene-co-acrylic acid), whose barrier properties are essential in the protection of the iron steel against corrosion. However, sodium hypochlorite is injected at certain points of the distribution network to ensure water potability, which can significantly limit the coating's ability to protect the iron steel and, therefore, imposes a lifetime. This thesis studies the aging mechanisms of this coating in a chlorinated water. First, it demonstrates the role of the copolymer’s acids on its physico-chemical and thermal properties. The formation of hydrogen bonds between the acids, gathered in clusters, reduces chain mobility and also imposes a distribution of a phenolic antioxidants between the ethylene phase and the clusters. The antioxidants distributed around the clusters are deactivated, so the protective capacity of the copolymer is reduced. It is shown that, during aging in chlorinated water, some of the antioxidants move to the acids, and another part is consumed by chlorinated species. This leads to a progressive destabilization of the copolymer. Finally, along with the loss of antioxidants, a reduction in the adhesion of the coating exposed to chlorine was observed due to the presence of water at the interface. As a result, the lifetime life has been estimated. To go further, a more in depth analysis of the copolymer’s oxidation mechanism could be done.Saint-Gobain Pont-à-Mousson produit des canalisations en fonte ductile destinées aux réseaux d’eau potable. Ces tuyaux sont revêtus intérieurement d’un copolymère thermoplastique, le poly(éthylène-co-acide acrylique), pour lequel les propriétés barrière sont indispensables dans la protection de la fonte contre la corrosion. Or, en certains points du réseau de distribution, l’hypochlorite de sodium est injecté afin d’assurer la potabilité de l’eau, ce qui peut limiter fortement la capacité du revêtement à protéger la fonte et impose donc une durée d’utilisation. Cette thèse étudie alors les mécanismes de vieillissement de ce revêtement en eau chlorée. Il est d’abord démontré le rôle fondamental des acides sur les propriétés physico-chimiques et thermiques du copolymère. La réalisation de liaisons hydrogène entre les acides conduit à la formation de clusters qui réduisent la perte de mobilité des chaînes en plus d’imposer une distribution des antioxydants phénoliques entre les phases éthylène et les clusters. Les antioxydants distribués autour des clusters sont désactivés affaiblissant la capacité de protection du copolymère. Il est démontré qu’au cours du vieillissement en eau chlorée, une partie des antioxydants est transférée vers les acides et une seconde partie est consommée par les espèces chlorées. L’ensemble conduit à une déstabilisation progressive du copolymère. Enfin, en parallèle de la perte d’antioxydants, une diminution de l’adhésion du revêtement exposé au chlore a été observée, qui serait due à la présence d’eau à l’interface. Ainsi, la durée d’utilisation a en partie pu être estimée. Cette étude pourrait être complétée par une analyse plus approfondie du mécanisme d’oxydation du copolymère
A bi-harmonic five-phase SPM machine with low ripple torque for marine propulsion
No full textInternational audienceThis paper addresses the design of a bi-harmonic five-phase Surface-mounted Permanent Magnet (SPM) machine for marine propulsion. The bi-harmonic characteristic results from the particular 20 slots-8 poles configuration that makes possible high value of third harmonic current injection. Thus the machine performance can be improved in terms of average torque, speed range, losses control and torque quality, this last feature being the scope of the paper. As low ripple torques are wanted at low speed, the magnet layer is defined to reduce the cogging torque and to make third harmonic current injection increasing average torque and reducing pulsating torque in the same time. According to a selection procedure based on the numerical simulations of a high number of machines, it appears that designing the rotor with two identical radially magnetized magnet that cover two-third the pole arc allows to reach this goal. Referring to an equivalent three-phase machine, the torque ripple level of the bi-harmonic five-phase machine is more than three times lower, thus being obtained with a simple control strategy that aims at achieving constant currents in the rotating frames. The time simulations of the drive confirm the significant reduction of the speed oscillation, especially at low speed
Energy-Based Model Representation for Design and Optimization of Piezoelectric Energy Harvesting Systems
No full textThe authorsacknowledge IRCICA (Research Institute onsoftware and hardware devices for informationand Advanced communication, USR CNRS3380) for providing equipment and facilities thatwere essential in advancing the project. The authors also wish to acknowledge the valuablecontributions of Frederic Giraud, Anis Kaci,Christophe Giraud-Audine, Clément Maillet,Cédric Lapeyronie and Sophie Barrau.International audienceThis paper presents a modular and adaptivemodelling and simulation tool for piezoelectricenergy harvesting (PEH) systems, applicable toportable micro-systems, such as wearabledevices, sensors, or bio-medical devices. Themain distinguishing feature of this tool is that it isdesigned to optimize the design phase and theenergy management strategies, thanks to asystemic approach. By employing EnergeticMacroscopic Representation (EMR) andinversion-based control, the tool facilitatesflexible analysis across varying levels of detail.Using this tool, a validated model is constructed,enabling the testing of power converter strategiesthat connect piezoelectric elements to loads, aswell as the evaluation of two energy harvestingstrategies for imposed displacement systems.This work demonstrates the potential of advancedmodelling techniques to enhance the design andimplementation of piezoelectric energy recoverysystem
Power Hardware and Driver-in-the-Loop for Battery Testing applied to Electrified Vehicles
No full textInternational audienceThis paper presents a unified approach to include a driver and a power subsystem in a testing simulation loop. The aim is to create a kind of lab-based vehicle offering an alternative to the Vehicle-in-the-Loop (ViL) approach or vehicle prototype for battery testing. This lab-based vehicle is built around experimental test benches and a driving simulator to reproduce a vehicle in a laboratory. A unified approach is used to easily identify inputs and outputs between the model vehicle and the emulation interfaces. The unified approach is based on the Energetic Macroscopic Representation (EMR) formalism. A Power Hardware-in-the-Loop (P-HiL) simulation is introduced for battery testing. Furthermore, standalone battery testing is extended to include a Driver in the Loop (DiL). A case study based on a retrofitted Plug-in Hybrid Electric Vehicle (P-HEV) is presented with experimental results
An Energetic Approach for Comparing Monomaterial Compliant Bistable Mechanisms
No full textInternational audienceShape-morphing systems offer multiple functionalities in a single part by leveraging compliance and bistability principles. Compliant mechanisms, composed of flexible structural elements, derive motion from deflection rather than from traditional joints. Structures exhibiting two stable equilibria are termed bistable. These bistable mechanisms can replace complex rigid body assemblies, enhancing modularity. However, their application is challenging due to the lack of comprehensive methods for comparing mechanical behaviors qualitatively and quantitatively. This research introduces a novel methodology for investigating the energetic properties and actuation symmetry of bistable mechanisms. Utilizing systematic review and meta-analysis, the study categorizes a dataset of articles into two classes, providing a robust reference for studying bistable mechanisms. The analysis focuses on how critical parameters such as motion type and shape affect behavior and actuation symmetry, using load–displacement curves and related energy metrics. The findings present a new method to identify key parameters and offer valuable design guidelines for developing compliant, single-part, and sustainable mechanisms