HAL Portal UBS (Université Bretagne Sud)
Not a member yet
14201 research outputs found
Sort by
Metabolomic fingerprint of Reynoutria japonica (Polygonaceae) using suspect screening with LC-HRMS and untargeted approach with GC-HRMS
No full textInternational audienc
Une approche de la reconstruction d’images et de la décomposition de matériaux en tomodensitométrie spectrale avec régularisation par modèle de diffusion
No full textSpectral computed tomography (CT) leverages the energy-dependent nature of X-ray attenuation to reconstruct material-specific images of the scanned object or patient. This reconstruction task constitutes an ill-posed inverse problem, requiring the incorporation of regularization or designing a prior.Recently, generative models, and particularly diffusion models, have shown remarkable performance in modeling natural image distributions. These developments have opened new possibilities for incorporating learned priors into inverse problems.This thesis investigates the use of diffusion models as learned priors for spectral CT reconstruction, where the strong correlation between energy channels is implicitly captured by the generative model prior. Furthermore, we extend this approach to material decomposition and propose a method that handles additional materials not present during training. We demonstrate that these models serve as powerful and flexible regularizers, outperforming classical techniques in terms of reconstruction quality.Les méthodes de tomographie spectrale exploitent la dépendance en énergie de l’atténuation des rayons X pour reconstruire des images spécifiques aux matériaux présents dans l’objet ou le patient scanné. Cette tâche de reconstruction constitue un problème inverse mal posé, nécessitant l’introduction de régularisation ou de connaissances a priori.Récemment, les modèles génératifs, et en particulier les modèles de diffusion, ont montré des performances remarquables dans la modélisation des distributions d’images naturelles. Ces avancées ouvrent de nouvelles perspectives pour l’intégration de priors appris dans des problèmes inverses.Cette thèse explore l’utilisation des modèles de diffusion comme a priori appris pour la reconstruction CT spectrale où la forte dépendance entre les canaux d'énergie est implicitement restituée par l'a priori du modèle génératif. De plus, nous étendons cette approche à la décomposition en matériaux et proposons une méthode dans le cas de matériaux additionnels non présents lors de l'apprentissage. Nous montrons que ces modèles constituent des régularisateurs puissants et flexibles, surpassant les techniques classiques en termes de qualité de reconstruction
Des liaisons élastiques transitoires au frottement : étude complète d'une équation du quatrième ordre pénalisée avec retard
No full textInternational audienceIn this paper we consider a fourth order nonlinear parabolic delayed problem modeling a quasi-instantaneous turn-over of linkages in the context of cell-motility. The model depends on a small parameter ε which represents a typical time scale of the memory effect. We first prove global existence and uniqueness of solutions for ε fixed. This is achieved by combining suitable fixed-point and energy arguments and by uncovering a nonlocal in time, conserved integral quantity. After giving a complete classification of steady states in terms of elliptic functions, we next show that every solution converges to a steady state as . When , we then establish convergence results on finite time intervals, showing that the solution tends in a suitable sense towards the solution of a parabolic problem without delay. Moreover, we establish the convergence of energies as , which enables us to show that, for ε small enough, the -dependent problem inherits part of the large time asymptotics of the limiting parabolic problem
Impact of pre-softening on rubber adhesion
No full textInternational audienceThe effect of stress softening, namely the Mullins effect, on elastomers' mechanical properties such as fatigue, strength, or toughness is widely studied, but mainly on bulk materials. The present contribution presents experimental evidence that applying a pre-softening load may significantly alter the critical strain energy release rate (SERR), which controls the interface adhesion between a soft solid and a rigid substrate. Here, we focus on the adhesion of a natural rubber matrix to a single metal wire, characterized using the Rubber Cord Adhesion Inflation Test. This protocol allows for the observation of a stable steady-state crack propagation under an axisymmetric configuration. An experimental protocol is proposed that utilizes a pre-load sequence to achieve a controlled softening of the rubber sheath surrounding the wire. Quantification is carried out through numerical simulations and subsequent material characterization. The adhesion results indicate that the pre-loading sequence reduces the specimen's overall resistance to decohesion. A more detailed analysis reveals that the critical SERR value is also lower. These findings provide new insights into the role of energy dissipation, both within the bulk material and at the interface level, in influencing the interface failure between an elastomer matrix and a rigid reinforcement
Adaptive fuzzy direct flux-vector control for stand-alone DFIG-AC systems improving voltage/frequency stability with battery storage
No full textInternational audienceThis paper proposes performances improvement of Direct Flux-Vector Control (DFC) applied to stand-alone Doubly Fed Induction Generator (DFIG). In a stand-alone AC power supply application, magnitude and frequency of the generated voltage need to be carefully monitored. With the DFC method, two closed loops are used to regulate the amplitude/frequency of the stator voltage via the amplitude/angle of the rotor flux, respectively. In this strategy, no position sensor or reference frame rotation are required, since it is carried out within the reference framework linked to the rotor. In addition, the rotor flux vector is adjusted by a direct selection of an appropriate voltage vector, making control structure implementation simple and easy. However, the DFC suffers from significant steady-state oscillations due to the presence of hysteresis regulators. To remedy this problem, a fuzzy inference control is introduced to enhance the efficiency of conventional DFC control by eliminating the hysteresis controllers and classical switching table. Indeed, intelligent identification of the proper voltage vector by the fuzzy DFC (F-DFC) seems to be a good solution for mitigating large ripples in rotor flux amplitude and angle. Furthermore, a fuzzy direct power control (F-DPC) is used to reduce active/reactive power ripples produced by the load-side converter. To ensure better power stability and improve energy management, a battery storage system is integrated, allowing compensation of power imbalances caused by wind speed and load variations. The effectiveness of the proposed control system is evaluated by multiple simulation tests performed with Matlab/Simulink software. The obtained results attest the capability of the suggested control structure to effectively regulate stator voltage amplitude and frequency under load and wind speed variation. Moreover, the findings confirm the superiority of the proposed F-DFC and F-DPC methods in terms of ripples minimization compared to C-DFC and C-DPC, respectively. To check the experimental feasibility of the recommended control technique, a real-time simulation is executed on the OP4510 system. The obtained experimental results validate those found by simulation and attest to the feasibility of the proposed F-DFC/F-DPC techniques for off-grid wind systems based on DFIG.</div
Santé psychologique des oncologues : déterminants et conséquences dans les soins
No full textInternational audienceIntroduction. While many publications have been made on burnout of oncologists, few publications have been made on oncologists in France, and none focuses on positive markers of their psychological health. Our study is a national study that studies the role of exercise mode, determinants and consequences of positive markers of psychological health of oncologists in France.Methodology. The survey was disseminated to oncologists (n = 606) by the learned societies or cooperative groups of the different cancer specialties from May to July 2022. The questionnaire assessed relationships between exercise mode, determinants and consequences for work, oncologist health and positive markers of psychological health.Results. French oncologists have high scores for positive markers of psychological health studied with differences according to speciality, status and mode of exercise. The determinants are the same, with inverse relationships, as those already known for burnout: negative impacts of the pro/personal life balance, perceived stress, workaholism, workload, professional isolation, positive impact of job recognition and management support. Important results concern therelationship of the factors studied with the health of oncologists, work behaviours but also in the relationship with patients (significant impact on empathy and compassion in care). Conclusion. Our study highlights the managerial and organisational factors that need to be worked on in order to preserve the psychological health of oncologists.Introduction. Si beaucoup de publications ont été faites sur le burn-out des oncologues, peu de publications ont été faites sur les oncologues en France et aucune ne s'attache aux marqueurs positifs de leur santé psychologique. Notre étude est une étude nationale qui étudie le rôle du mode d'exercice, les déterminants et les conséquences de marqueurs positifs de la santé psychologique des oncologues en France.Méthodologie. L'enquête a été diffusée auprès des oncologues (n = 606) par les sociétés savantes ou groupes coopérateurs des différentes spécialités de cancérologie de mai à juillet 2022. Le questionnaire évaluait les relations entre le mode d'exercice, les déterminants et les conséquences sur le travail, la santé des oncologues et les marqueurs positifs de la santé psychologique.Résultats. Les oncologues français présentent des scores élevés pour les marqueurs positifs de santé psychologique étudiés avec des différences selon la spécialité, le statut et le mode d'exercice. Concernant les déterminants, on retrouve les mêmes, avec des relations inverses, que ceux déjà connus pour le burn-out : impacts négatifs de l'équilibre vie pro/vie perso, du stress perçu, du workaholisme, de la charge de travail, de l'isolement professionnel, impact positif de lareconnaissance au travail et du soutien de la direction. Des résultats importants concernent la relation des facteurs étudiés avec la santé des oncologues, les comportements au travail mais aussi dans la relation avec les patients (impact important sur l'empathie et la compassion dans les soins).Conclusion. Notre étude met en évidence les facteurs managériaux et organisationnels devant être travaillés pour préserver la santé psychologique des oncologues
Optimizing reliability and safety of wind turbine systems through a hybrid control technique for low-voltage ride-through capability
No full textInternational audienceThis study addresses a significant challenge in reliability engineering and system safety, specifically the operation of wind turbines under fault conditions. It proposes an asymmetrical fault ride-through (AFRT) control method designed for the doubly fed induction generator (DFIG) rotor-side converter (RSC) used in wind turbines. The DFIG model is analyzed in both positive and negative rotating synchronous reference frames (PR-SRF and NR-SRF), incorporating four key components to prevent overcurrent in the RSC during AFRT conditions. The proposed control method is divided into two segments: first, reducing the four components based on boundary constraints and reference value configuration; and second, determining the control characteristic ‘ k ’ through an optimization loop using the particle swarm optimization (PSO) algorithm. The effectiveness of the PSO algorithm is compared with three other optimization methods genetic algorithm (GA), differential evolution (DE), and ant colony optimization (ACO). The dynamic performance of the proposed method is assessed under Line-to-Line (LL) and Line-to-Line-to-Ground (LLG) fault scenarios. Simulation results demonstrate that the method successfully mitigates fluctuations caused by asymmetrical faults (AFs), achieving a 7.2% higher efficiency in AFRT than similar approaches. Ultimately, this research enhances wind turbine system safety and reliability, ensuring more robust power generation during asymmetrical fault conditions
Starvation, Participation, and Exclusion. Friedrich Justus Riedel and the Bread Riot of Erfurt, 1771
No full textInternational audienceIn 1771, as Germany was experiencing one of the worst famines in its history, the people of Erfurt turned against the city’s mayor, Siegmund Lebrecht Hadelich. The mayor, who was also a professor of cameral sciences at the university, was accused of hoarding grain, speculating on the shortage and publicly advocating the virtues of free trade. This upheaval provides an opportunity to analyse how the demand for the fundamental right to food has led to a dual process of attempts to expand and contract political participation rights. Surprisingly, the rioters received unexpected support from another university professor, the aesthetic philosopher Friedrich Justus Riedel (1742–1785). For Riedel, the riot was an opportunity to examine the place of the common man in society, his right to speak, and his political competence. This article summarizes the initial results of a project carried out as part of a French habilitation à diriger les recherches
Rational design of mesoporous NiWO4 / Co3O4/ g-C3N4 based heterostructure for high performance asymmetric supercapacitors
No full textInternational audienceHerein we have designed a nickel-based tungsten oxide as a cathode material for hybrid supercapacitor owing to its better theoretical capacitance values. However, the material lacks good conducting behaviour and rate capability which affect its extensive utilisation as electrode material. In order to overcome the defects heterostructure composite was designed to enhance its electrochemical behaviour. The prepared materials were characterised with several physiochemical and electrochemical techniques. The ternary nanocomposite has portrayed cube like morphology with mesoporous nature. In specific, the ternary composite in three electrode cell have delivered a high specific capacitance of 819 F/g at 1 A/g when compared to other materials and retained a initial capacitance of 91 % after 5000 charge discharge cycles. Then an asymmetric hybrid supercapacitor with NiWO4/Co3O4/g-C3N4 as positive electrode and rGO as negative electrode in 3 M PVA-KOH electrolyte using Swagelok cell was assembled. The fabricated device exhibited a specific capacitance of about 113 F/g at 1 A/g in the potential of 1.4V with a specific energy of 35 Wh/kg at an specific power of 1500 W/kg. Further the device exhibited a better cycle life of 93 % even after 10000 cycles. Thus, the transition metal tungstate based heterostructure could be employed as a potential electrode material for efficient supercapacitors