HAL Portal UTC Université de Technologie de Compiègne
Not a member yet
11652 research outputs found
Sort by
Sustainability of water transfers in the Crau plain
No full textInternational audienceThe Crau plain, 600 km2 located in South-East France, is mainly associated with the production of high quality hay (around 15,000 ha) irrigated from open-channel networks. Traditional irrigation practices consist in high discharge in order to reach the end of long plots, the excess of water being both drained by run-off to ditches and percolated to the so-called “Crau aquifer”. The aquifer recharge depends for around 70% on hay irrigation, the organization of its management thus relies on the sustainability of irrigation practices. However, hay production faces social and physical pressures from local to regional scales. Socially, water management in the fields requires to be fine-tuned to balance working time dedicated to irrigation (difficult labor conditions with high workload and night shifts) with water flows throughout irrigated plots, farms and canals. Physically, the low-performance hay irrigation is under tension because of local land-use changes due to the development of urban areas and other agricultural production (orchards and horticulture), in a context of hydraulic infrastructures requiring important rehabilitation works. Locally, return flows provide a mix of interdependent services, the aquifer being used for the extraction of drinking water for 300,000 inhabitants, for other irrigated crops like orchards, and for industries. Regionally, water comes from an historical inter-basin transfer, passing through a succession of hydraulic infrastructures and hydroelectric power plant before entering the plain. The climate change impacts on upstream precipitation make incoming water being less abundant, leading to water restrictions as experienced in 2022. The sustainability of water transfers questions the integration of land and water planning. The aim of our research is to propose an original perspective coupling the characterization of water flows in relation to irrigation practices at the plot and scheme scales with the evaluation of farmers leeway in terms of economic and organizational constraints. The objective of this communication is to present each part of this work and to draw up further correspondences between the hydraulic and economic dimensions. First, an agrarian diagnosis revealed the lack of information on water flows, motivating in turn the original development of affordable measuring devices to track water in an irrigated block and automate parts of irrigation practices. Second, the context of water and land increasing scarcities motivated the characterization of the vulnerability of hay productions in terms of access to water, labor and markets. These studies aimed to directly contribute to water management in the Crau plain, respectively in the search for technical optimization to use water in the agricultural system more efficiently (contributing to reduce working flows) and for the definition and evaluation of strategies for adapting agriculture to meet the challenges of farm economics, groundwater recharge and water conservation. Finally, we will draw on both inputs to assess land cover scenarios and their impacts on aquifer recharge; we may also evaluate possible impacts of water restrictions on land uses
A New Energetic Reasoning for the Cumulative Scheduling Problem: A Dynamic Programming Approach
No full textInternational audienc
Dual-Level Control Architectures for Over-Actuated Autonomous Vehicle's Stability, Path-Tracking, and Energy Economy
No full textInternational audienceAutonomous vehicles equipped with four independent in-wheel motors bestow salutary design flexibility and render the system over-actuated. The strategy percolated for torque allocation dictates the system's performance and marks its energy consumption. In this paper, two complete novel control architectures are developed and contrasted from the viewpoint of vehicle performance and energy consumption. A cascaded control strategy is employed by incorporating two distinct control levels. The high level is differentiated by a centralized approach based on the optimal H∞ control in the framework of the Linear Parameter Varying (LPV) systems, and a decentralized approach based on problem decoupling where a solution is proposed using the Super-Twisting Sliding Mode (STSM) control. Both approaches are supervised by a decision layer to promote the stability objective in critical driving situations. At the low level, stability control based on Direct Yaw Control (DYC) along with speed control are both achieved using an original torque allocation strategy. A comprehensive set of four multi-objective strategies has been devised, centered around a proposed torque allocation configuration. These strategies encompass dynamic online optimization, expertly solved using the highly efficient Sequential Quadratic Programming (SQP) method, as well as a unique offline optimization based on a data-driven implemented algorithm. The proposed architectures are tested and validated in a joint simulation between Simulink/MatLab and SCANeR TM Studio vehicle dynamics simulator. The simulation findings demonstrate substantial improvements in stability, comfort, and energy efficiency at both the high and low levels of the autonomous in-wheel-driven electric vehicle
Uncertainty Quantification in Logistic Regression using Random Fuzzy Sets and Belief Functions
No full textInternational audienceEvidential likelihood-based inference is a new approach to statistical inference in which the relative likelihood function is interpreted as a possibility distribution. By expressing new data as a function of the parameter and a random variable with known probability distribution, one then defines a random fuzzy set and an associated predictive belief function representing uncertain knowledge about future observations. In this paper, this approach is applied to binomial and multinomial regression. In the binomial case, the predictive belief function can be computed by numerically integrating the possibility distribution of the posterior probability. In the multinomial case, the solution is obtained by a combination of constrained nonlinear optimization and Monte Carlo simulation. In both cases, computations can be considerably simplified using a normal approximation to the relative likelihood. Numerical experiments show that decision rules based on predictive belief functions make it possible to reach lower error rates for different rejection rates, as compared to decisions based on posterior probabilities
Synergies between machine learning and reasoning - An introduction by the Kay R. Amel group
No full textInternational audienceThis paper proposes a tentative and original survey of meeting points between Knowledge Representation and Reasoning (KRR) and Machine Learning (ML), two areas which have been developed quite separately in the last four decades. First, some common concerns are identified and discussed such as the types of representation used, the roles of knowledge and data, the lack or the excess of information, or the need for explanations and causal understanding. Then, the survey is organised in seven sections covering most of the territory where KRR and ML meet. We start with a section dealing with prototypical approaches from the literature on learning and reasoning: Inductive Logic Programming, Statistical Relational Learning, and Neurosymbolic AI, where ideas from rule-based reasoning are combined with ML. Then we focus on the use of various forms of background knowledge in learning, ranging from additional regularisation terms in loss functions, to the problem of aligning symbolic and vector space representations, or the use of knowledge graphs for learning. Then, the next section describes how KRR notions may benefit to learning tasks. For instance, constraints can be used as in declarative data mining for influencing the learned patterns; or semantic features are exploited in low-shot learning to compensate for the lack of data; or yet we can take advantage of analogies for learning purposes. Conversely, another section investigates how ML methods may serve KRR goals. For instance, one may learn special kinds of rules such as default rules, fuzzy rules or threshold rules, or special types of information such as constraints, or preferences. The section also covers formal concept analysis and rough sets-based methods. Yet another section reviews various interactions between Automated Reasoning and ML, such as the use of ML methods in SAT solving to make reasoning faster. Then a section deals with works related to model accountability, including explainability and interpretability, fairness and robustness. Finally, a section covers works on handling imperfect or incomplete data, including the problem of learning from uncertain or coarse data, the use of belief functions for regression, a revision-based view of the EM algorithm, the use of possibility theory in statistics, or the learning of imprecise models. This paper thus aims at a better mutual understanding of research in KRR and ML, and how they can cooperate. The paper is completed by an abundant bibliography
Voir le son : Représentations visuelles de la musique dans l'iconographie médiévale à l'aide de la base de données Musiconis
No full textInternational audienceThis article explores how medieval visual art represents musical practices, us ing data from the Musiconis database—a structured, searchable corpus of over 3,000 images from the 9th- to 16th-centuries. These images, found in sculptures, manuscripts, stained glass, and other media, offer rich insight into the instru ments, performers, and contexts of medieval music. The database provides detailed metadata, enabling large-scale quantitative and qualitative analysis of iconographic trends, including instrument types, performer roles, gender, and geographic origins. Aerophones and chordophones dominate the visual record, while idiophones and membranophones are less represented. Performer typologies reveal a growing diversity over time: instrumentalists remain prevalent, but dancers, singers, acrobats, and assistants appear increasingly from the 13th-century onward. Gendered representations, especially of female musicians, highlight both allegorical and historical roles. Musical functions range from liturgical and angelic to courtly, parodic, erotic, and even diabolic, illustrating sound’s symbolic and spiritual significance. The study also shows regional and temporal biases—France is overrepresented, and earlier centuries are sparsely documented. Analytical tools like Python and Colab enabled dynamic visualizations of trends in musical iconography. Case studies, such as the rise of the lute and decline of the rota, reveal shifts in instrumental use and performance practice. The article concludes by proposing future directions, including expanding the corpus, integrating Mediterranean and ancient sources, and applying phylogenetic models to trace organological evolution.Cet article explore la manière dont l'art visuel médiéval représente les pratiques musicales, à partir des données de la base de données Musiconis – un corpus structuré et interrogeable de plus de 3 000 images datant du IXe au XVIe siècle. Ces images, présentes dans des sculptures, des manuscrits, des vitraux et d'autres supports, offrent un éclairage précieux sur les instruments, les interprètes et le contexte de la musique médiévale. La base de données fournit des métadonnées détaillées, permettant une analyse quantitative et qualitative à grande échelle des tendances iconographiques, notamment les types d'instruments, les rôles des interprètes, le genre et les origines géographiques. Les aérophones et les chordophones dominent les représentations visuelles, tandis que les idiophones et les membranophones sont moins fréquents. La typologie des interprètes révèle une diversité croissante au fil du temps : les instrumentistes restent prédominants, mais les danseurs, les chanteurs, les acrobates et les assistants apparaissent de plus en plus à partir du XIIIe siècle. Les représentations genrées, en particulier celles des musiciennes, mettent en lumière des rôles à la fois allégoriques et historiques. Les fonctions musicales, allant du liturgique et angélique au courtois, au parodique, à l'érotique, voire au diabolique, illustrent la portée symbolique et spirituelle du son. L'étude révèle également des biais régionaux et temporels : la France est surreprésentée et les siècles anciens sont peu documentés. Des outils analytiques tels que Python et Colab ont permis des visualisations dynamiques des tendances de l'iconographie musicale. Des études de cas, comme l'essor du luth et le déclin de la rota, mettent en lumière des changements dans l'usage instrumental et les pratiques d'interprétation. L'article conclut en proposant des pistes de recherche futures, notamment l'élargissement du corpus, l'intégration de sources méditerranéennes et antiques, et l'application de modèles phylogénétiques pour retracer l'évolution organologique.Questo articolo esplora il modo in cui l’arte visiva medievale rappresenta le pratiche musicali, utilizzando i dati del database Musiconis, un corpus strutturato e consultabile di oltre 3.000 immagini dal IX al XVI secolo. Queste immagini, presenti in sculture, manoscritti, vetrate e altri supporti, offrono una ricca panoramica sugli strumenti, gli esecutori e i contesti della musica medievale. Il database fornisce metadati dettagliati, consentendo un’analisi quantitativa e qualitativa su larga scala delle tendenze iconografiche, inclusi tipi di strumenti, ruoli degli esecutori, genere e origini geografiche. Aerofoni e cordofoni dominano la documentazione visiva, mentre idiofoni e membranofoni sono meno rappresentati. Le tipologie di esecutori rivelano una crescente diversità nel tempo: gli strumentisti rimangono prevalenti, ma danzatori, cantanti, acrobati e assistenti compaiono sempre più spesso dal XIII secolo in poi. Le rappresentazioni di genere, soprattutto delle musiciste, evidenziano ruoli sia allegorici che storici.Le funzioni musicali spaziano da quelle liturgiche e angeliche a quelle cortesi, parodistiche, erotiche e persino diaboliche, illustrando il significato simbolico e spirituale del suono. Lo studio evidenzia anche distorsioni regionali e temporali: la Francia è sovrarappresentata e i secoli precedenti sono scarsamente documentati. Strumenti analitici come Python e Colab hanno consentito visualizzazioni dinamiche delle tendenze nell’iconografia musicale. Casi di studio, come l’ascesa del liuto e il declino della rota, rivelano cambiamenti nell’uso strumentale e nella pratica esecutiva. L’articolo si conclude proponendo direzioni future, tra cui l’espansione del corpus, l’integrazione di fonti mediterranee e antiche e l’applicazione di modelli filogenetici per tracciare l’evoluzione organologica
Efficient Time Slot Recovery Protocol for Wireless Networks
No full textInternational audienceDense and resource-constrained wireless networks, such as the Internet of Things (IoT), pose significant challenges for Medium Access Control (MAC) protocols in addressing communication collisions and Quality of Service (QoS) issues. Interference and message collisions within communication channels are primary culprits behind network faults and energy wastage. Recent research has explored methods for enabling large-scale networks to communicate efficiently, utilizing local information and avoiding collisions through distributed and parallel distance-2 TDMA (Time-Division Multiple Access) coloring. However, a critical aspect that has been largely unexplored is how highly active nodes can utilize the time slots of less active neighbors without message exchanges, thereby enhancing network performance, especially in broadcasting emergency messages. In this paper, we introduce a distributed protocol designed to enhance communication and energy efficiency in wireless networks by effectively utilizing underused time slots from neighboring nodes. Our protocol enables more active nodes to efficiently recover these underused time slots, facilitating the rapid broadcast of urgent messages. We conclude by presenting and discussing experimental results to demonstrate the efficacy of our proposed solution
Faire campagne à l'ère du numérique : quelles participations pour quelles dynamiques d'influence des organisations de plaidoyer ? : analyse au prisme d'une position embarquée au sein de Greenpeace France
No full textOften touted as a revolutionary tool for enhancing the visibility of a cause, the deployment of digital technology within an organization is more complicated than that. To make the most of the benefits of digital information and communication technologies, it is imperative to adapt to the new operational modalities of the public sphere and the participation culture inherent to these spaces. Building on this premise, this thesis aims to examine the shifts precipitated by the integration of digital technology into advocacy. It unpacks “the place of participation,” focusing on a central question: under what conditions can online participation in support of an advocacy organization generate influence? To better understand the influence of participation, this research adopts an embedded approach centered in observing how campaign actors use digital tools. Positing that it is possible to develop tools with more "top-down" or "bottom-up" orientations, this thesis seeks to demonstrate that these orientations elicit forms of participation that impact visibility and connectivity. The form of participation influences advocacy on two levels. Firstly, it has an impact on the “issue network” that forms around the object of a campaign, and in which participation can help define the agenda and frame the public issue. Secondly, it feeds the “network internalities” of a campaign, and this can be seen through participation dynamics that reinforce the resilience of organizations (resources and reputation) and participants (activist and civic skills). To test this theoretical framework, I was embedded in three campaigns of Greenpeace France as a "doctoral student-practitioner-researcher”. To study the campaigns, I combined ethnographic work and semi-structured interviews with relevant staff members. To assess how their work was received by the public, I used digital methods and qualitative tools (questionnaires, focus groups, and interviews). Through these three case studies, I demonstrated that the influence of online participation can only be fully harnessed if it manages to both achieve visibility (beyond an activist audience) and connect with the public (beyond clicks). Faced with the organizational challenge of generating influential participation, I identified a more pragmatic campaigning approach that adapts to the communicative constraints of online platforms. This approach consists of designing campaigns truly centered on the public, who can meaningfully contribute to achieving a predefined objective.Souvent vanté comme un outil révolutionnaire pour renforcer la visibilité d’une cause, le déploiement du numérique dans une organisation est plus complexe. Pour espérer tirer profit des technologies numériques d’information et de communication, il est impératif de s’adapter aux nouvelles modalités de fonctionnement de l’espace public et à la culture de la participation caractéristique de ces espaces. Partant de ce constat, cette thèse a pour objectif de rendre compte des transformations générées par l’intégration du numérique à la pratique d’influence qu’est le plaidoyer. Elle questionne « la juste place » de la participation, en se focalisant sur une problématique centrale : à quelles conditions la participation en ligne en soutien d’une organisation de plaidoyer peut être génératrice d’influence ? Pour saisir cette influence de la participation, cette recherche se fonde sur une approche environnée, centrée sur l’observation des usages du numérique par les acteurs d’une campagne. En postulant qu’il est possible de construire des dispositifs à l’orientation plus ou moins « top-down » ou « bottom-up », cette thèse entend démontrer que ces orientations suscitent des formes de participations plus ou moins visibles et connectées. Ces formes de participation exercent une influence à deux niveaux. Le premier niveau s’exerce sur un réseau d’action thématique qui se forme autour de l’objet d’une campagne et où la participation peut contribuer à définir l’agenda et cadrer le problème public. Le deuxième niveau vient nourrir le réseau interne d’une campagne, s’observant à travers les dynamiques de participation qui viennent renforcer la résilience d’une organisation (ressources et notoriété) et des participants (compétences militantes et civiques). Pour tester ce cadre théorique, j’ai embarqué au sein de trois campagnes de Greenpeace France en tant que « doctorant-praticien-chercheur-engagé ». Pour étudier leurs constructions j’ai combiné un travail ethnographique d’observation et des entretiens semi-directifs avec leurs acteurs. Pour examiner leurs réceptions auprès des publics, j’ai utilisé des méthodes numériques et des outils qualitatifs (questionnaires, focus groupes et entretiens). À travers ces trois cas d’étude, j’ai démontré que la participation en ligne peut seulement être pleinement influente si elle parvient à être visible (au-delà d’un public militant) et connectée (au-delà du clic). Face à la difficulté organisationnelle de générer cette participation d’influence, une approche plus pragmatique de faire campagne a été identifiée afin de s’adapter aux contraintes communicationnelles du web des plateformes. Celle-ci consiste à concevoir des campagnes réellement centrées sur les publics, capables de contribuer à atteindre un objectif prédéfini
Intensification du procédé d’extraction, de purification et de fonctionnalisation des acides caféoylquiniques à partir de coproduits de la culture de l’endive
No full textForced chicory root is a by-product of Belgian endive culture, a typical crop of northern France, Belgium and the Netherlands. Currently under-utilized in methanation or animal feed, this by-product contains molecules of interest: caffeoylquinic acids. These molecules have antioxidant and anti-inflammatory properties, and a potential for reducing metabolic disorders. This thesis aims to intensify the pre-treatment, extraction, purification and functionalization of caffeoylquinic acids from forced chicory roots to develop new bioactive biosourced molecules of potential interest to the cosmetics and nutraceutical sectors. The final part of the thesis deals with a technico-economical study of the process to estimate its economic profitability in relation to the targeted application sector. The first part focuses on the effect of conventional pretreatments (cutting and drying) and the effect of pulsed electric field pretreatment on caffeoylquinic acid content in biomass. The effect of adding an antioxidant solution during extraction is also investigated. Secondly, extraction optimization is carried out using dry or fresh biomass. The influence of factors such as temperature, solid/liquid ratio and solvent type were studied. In addition, extraction kinetics were performed to study kinetic parameters using empirical models. As the purity of the extract obtained is low, purification steps are needed. The thesis then focused on purifying the crude extract obtained using macroporous resins and liquid/liquid extraction. For resin purification, resin screening was carried out, followed by optimization of the purification operating conditions with the chosen resin. Models of adsorption phenomena are carried out to identify the limiting stages and the maximum adsorption capacity. For liquid/liquid extraction, green solvent screening is carried out on aqueous and hydro-ethanolic media, followed by optimization of operating conditions with the best solvent. The penultimate part of the thesis seeks to functionalize caffeoylquinic acids by esterification, starting with a model solution and then a real extract. Esterification conditions are optimized to increase both reaction speed and conversion rate. Esters with different chain lengths were obtained, and biological activities such as antioxidant activity and anti-UV properties were studied. Functionalization is also performed with real extract. A technico-economic study concludes the thesis, opening up prospects for the industrialization of the forced chicory roots valorization process.La racine forcée d’endive est un coproduit de la culture de l’endive, produit caractéristique du Nord de la France, de la Belgique et des Pays-Bas. Actuellement sous utilisé en méthanisation ou en alimentation animale, ce coproduit contient pourtant des molécules d’intérêts à haute valeur ajoutée : les acides caféoylquiniques. Ces molécules possèdent des activités antioxydantes, anti-inflammatoires et permettent de limiter les maladies du désordre métabolique. Ce travail de thèse vise à intensifier le prétraitement, l’extraction, la purification et la fonctionnalisation des acides caféoylquiniques à partir des racines forcées pour développer de nouvelles molécules bioactives biosourcées potentiellement intéressantes pour le secteur cosmétique et nutraceutique. Une dernière partie de la thèse porte sur l’étude technico-économique du procédé pour estimer sa rentabilité économique en fonction du secteur d’application visé. La première partie porte sur l’effet des prétraitements conventionnels (découpe et séchage) ainsi que l’effet d’un prétraitement électrique par champs électriques pulsés sur les teneurs en acides caféoylquiniques dans la biomasse. L’effet de l’ajout d’une solution antioxydante lors de l’extraction est également étudié. Dans un deuxième temps, une optimisation de l’extraction est réalisée à partir de biomasse sèche et fraiche. L’influence de facteurs tels que la température, le ratio solide/liquide, ainsi que la nature du solvant a été étudiée. De plus, des cinétiques d’extraction ont été tracées pour étudier les paramètres cinétiques à l’aide d’un modèle empirique. La pureté de l’extrait obtenu étant faible, des étapes de purification sont donc nécessaires. Par la suite, les travaux se sont portés sur la purification de l’extrait brut à l’aide de résines macroporeuses ainsi que par extraction liquide/liquide. Pour la purification par résine, un screening de résines est réalisé suivi d’une optimisation des conditions opératoires de purification avec la résine choisie. Des modélisations des phénomènes d’adsorption sont réalisées pour déterminer les étapes limitantes ainsi que la capacité maximale d’adsorption. Pour l’extraction liquide/liquide, un screening de solvants verts est effectué à partir d’un milieu aqueux et hydro-éthanolique puis une optimisation des conditions opératoires avec le meilleur solvant est réalisée. La pénultième partie de la thèse cherche à fonctionnaliser par estérification les acides caféoylquiniques à partir d’une solution modèle puis d’un extrait réel. Les conditions d’estérification sont optimisées pour augmenter la vitesse de réaction ainsi que le taux de conversion. Des esters avec différentes longueurs de chaine sont obtenus et l’activité antioxydante ainsi que les propriétés anti-UV sont étudiées. La fonctionnalisation est par la suite effectuée sur un extrait réel. Une étude technico-économique conclut la thèse permettant d’ouvrir sur des perspectives quant aux conditions nécessaires à l’industrialisation du procédé de valorisation des racines forcées d’endive
Modélisation numérique du comportement statique des stacks de tôles
No full textNational audienceL’objectif des travaux présentés est de proposer une démarche de modélisation du comportement mécanique d’un empilement de tôles minces utilisé dans la fabrication de machines électriques. Nous présentons les modèles de contact utilisés ainsi que leur implémentation éléments finis. Le modèle est ensuite appliqué sur un exemple de compression, soulignant l’effet de l’état de surface des tôles sur le comportement de l’assemblage. La modélisation proposée permet de reproduire le comportement normal de l’empilement en tenant compte du caractère irréversible dû à la plasticité locale