Dissertationen zum Thema „Jumeaux numériques“
Geben Sie eine Quelle nach APA, MLA, Chicago, Harvard und anderen Zitierweisen an
Machen Sie sich mit Top-22 Dissertationen für die Forschung zum Thema "Jumeaux numériques" bekannt.
Neben jedem Werk im Literaturverzeichnis ist die Option "Zur Bibliographie hinzufügen" verfügbar. Nutzen Sie sie, wird Ihre bibliographische Angabe des gewählten Werkes nach der nötigen Zitierweise (APA, MLA, Harvard, Chicago, Vancouver usw.) automatisch gestaltet.
Sie können auch den vollen Text der wissenschaftlichen Publikation im PDF-Format herunterladen und eine Online-Annotation der Arbeit lesen, wenn die relevanten Parameter in den Metadaten verfügbar sind.
Sehen Sie die Dissertationen für verschiedene Spezialgebieten durch und erstellen Sie Ihre Bibliographie auf korrekte Weise.
Hammoumi, Adam. „Analysis-Driven Design of Digital Multi-scale Microstructures of Materials“. Electronic Thesis or Diss., université Paris-Saclay, 2022. http://www.theses.fr/2022UPASG083.
Der volle Inhalt der QuelleThe global demand for energy is increasing rapidly around the world. Heterogeneous catalysis is behind most of the principles of green chemistry: energy saving processes, atom efficiency, cleaning processes, etc. Catalysis in general, and heterogeneous catalysis, has very quickly become an essential tool for the development of industrial chemistry. Currently, 85 % of the industrial processes used throughout the world are catalytic. In this context, many efforts are being made to optimize heterogeneous catalysts to meet the increasing demand for energy while reducing the environmental impact of fuels. For these reasons, and many others, the design of new catalytic materials today is a hot topic. The current approach regarding the design of catalysts tends towards a controlled elaboration of materials whose texture will be controlled, ordered, and hierarchically structured from the nanometer to the micrometer scale. These catalysts will have to be more active, and stable (energy efficiency), and more selective (saving atoms, less rejects). Within this framework, this research work focuses on the creation of numerical twins of microstructures aiming in-fine the retro-design of porous materials for optimal usage properties. The applications concern catalysis supports and construction materials, with priority given to the improvement of multi-physical properties considering textural properties of these materials. The contribution of this thesis is a new numerical framework allowing the modeling and characterization of these materials. This new approach builds upon random models to represent realistic multi-scale and complex microstructures. To extract the textural properties of these microstructures, two numerical methods have been developed in a first step. The first one allows us to characterize the porosity network of a microstructure and to compute its pore size distribution, and the second one allows us to compute the geometrical tortuosity of these materials with a fast graph search approach. Regarding the same topic, our major contribution is a morphological model that simulates in an original way gas physisorption by means of mathematical morphology and percolation operators. Gas physisorption is one of the most used experimental techniques for the characterization of the textural properties of porous materials. The model has been validated on real materials designed for this thesis. Deep learning has also been widely explored. First, a new approach building upon convolutional neural networks has been proposed. The latter proposes a solution to improve the learning quality when there is little input training data. A second contribution allowed us to store the morphological information of the previous model in a 3D volume, and to capture inter-slice information into 2D slices. The overall process transformed the initial problem into a deep learning problem, which considerably reduced the computation time of the gas physisorption model
Chabanet, Sylvain. „Contributions aux ombres et jumeaux numériques dans l’industrie : proposition d’une stratégie de couplage entre modèles de simulation et d’apprentissage automatique appliquée aux scieries“. Electronic Thesis or Diss., Université de Lorraine, 2023. http://www.theses.fr/2023LORR0131.
Der volle Inhalt der QuelleThis thesis is part of the ANR project Lorraine-Artificial Intelligence, a multi-disciplinary project promoting research into both artificial intelligence itself, and its applications to other fields. As such, this thesis focuses on the development and use of machine learning models as a substitute for simulation models. Interest in this research topic is fueled by academic and industrial interest in the concept of digital shadows and twins, seen as an evolution of simulation models for long-term use at the heart of systems and processes. The main contribution of this thesis is the proposal of a coupling strategy between a simulation model and a surrogate model performing the same prediction task repeatedly on a data stream. The simulation model is assumed to have a high level of fidelity, but to be too slow or computationally expensive to be used alone to perform the full range of prediction required. The surrogate model is a fast machine-learning model that approximates the simulation model. The primary objective of the proposed coupling strategy is the efficient use of limited computational resources by intelligently allocating each prediction request to one of the two models. This allocation is, in particular, inspired by active learning and based on the evaluation of the level of confidence in the predictions of the machine learning model. Numerical experiments are first carried out on eight datasets from the scientific literature. An application to the sawmilling industry is then developed
Haoua, Abdoulaye affadine. „Smart Machining pour l’assemblage aéronautique“. Electronic Thesis or Diss., Paris, HESAM, 2023. http://www.theses.fr/2023HESAE090.
Der volle Inhalt der QuelleHybrid stacks comprising of composite CFRP layers, TA6V, and aluminum alloy, referred to as multi-material, are widely used in aerospace components. One-shot drilling, using an Automatic Drilling Unit (UPA), of these stacks poses major challenges due to the difference in machinability among the various materials within the stack. To meet quality requirements, improve tool life and achieve optimal productivity during the drilling of multi-material stacks with an electric UPA, it is essential to adapt cutting parameters during machining. However, there are currently no tools available to meet this need. This theisis conerns machine learning-based solutions for the implementation of smart drilling strategies incorporating adaptive machining. These solutions focus on the development of robust and reliable methodologies for identifying materials, detecting tool wear, and detecting multi-material drilling-related incidents (tool breakage, chip jamming, lubrication issues, cutting edges chipping). Fristly, a new approach to incident detection was proposed, based on unlabeled industrial data and clustering algorithms: K-means and hierarchical ascending classification. This approach enabled us to identify a number of incidents, and highlighted the difficulties associated with processing industrial data (rarity of incidents and absence of labels) and the difficulties associated with defining the optimum parameters for machine learning algorithms. Subsequently, experiment are performed on an specialized instrumented drilling bench and on Al7175/CFP and CFRP/Al7175 stacks. This led to the development and validation of an original material identification methodology based on Random Forest (RF). The results, in terms of material classification (Al7175, CFRP), enabled the extension of the RF model to different stack configurations with various tool diameters and a wide range of cutting conditions. This methodology will enable the implementation of an adaptive machining process on an electric UPA for multi-material drilling.Keywords : Adaptive drilling, Multi-material stacks drilling, Material recognition, Random Forest, Tool Wear, Unsupervised learning, Electric UPA, Signals analysis
Leclerc, Lucas. „Quantum computing with Rydberg atoms : control and modelling for quantum simulation and practical algorithms“. Electronic Thesis or Diss., université Paris-Saclay, 2024. http://www.theses.fr/2024UPASP046.
Der volle Inhalt der QuelleRefining our understanding of an unknown system through modelling lays the groundwork for optimally controlling it and opens the door to a myriad of potential applications, exploiting the once enigmatic and unpredictable effects of this now-known system. This thesis applies this paradigm to analog quantum computing with Rydberg atoms, showcasing how careful noise modelling, optimal control and machine learning frameworks can support and enhance the simulation of quantum magnetism and the solving of graph-based optimisation and classification problems. After describing the experimental platform enabling the control of Rydberg atoms, we introduce classical tools such as digital twins of noisy systems, tensor network modelling, robust optimal control, and Bayesian optimisation for variational algorithms. We apply the latter to several applications. We improve the preparation of antiferromagnetic state in the Ising model and benchmark the noisy behaviour of a dipolar XY quantum simulator when probing continuous symmetry breaking and performing quantum state tomography. Using optimisation techniques and machine learning methods, we also tackle industrial use cases such as maximum independent set on graphs representing smart charging tasks, binary classification of toxic or harmless molecular compounds, and prediction of fallen angel companies in financial risk management
Coupry, Corentin. „Approche mixte RV/RA, couplée au concept de jumeau numérique, comme support d’opérations de maintenance collaborative à distance : application aux équipements techniques de bâtiment“. Electronic Thesis or Diss., Angers, 2023. http://www.theses.fr/2023ANGE0066.
Der volle Inhalt der QuelleTechnological advances in Industry 4.0 have paved the way for new methodologies and technologies, including digital twins (DT) and extended reality (XR) tools. While these innovations are already widely exploited in industrial maintenance, their potential in the construction sector remains to be demonstrated. They promise to help decision-making by providing access to specific information, via XR, during maintenance operations, and by enabling behavioral simulations thanks to the DT. However, the growing complexity of installations, combined with the need to call on external expertise during maintenance operations, calls for a rethinking of remote assistance methods, particularly in the light of environmental and health concerns. To explore these challenges, our research work, based on an in-depth review of scientific and technological literature, focuses on exploiting these technologies to improve maintenance procedures. Firstly, a literature review explored how XR tools and digital twins can enrich the information used during maintenance operations. This analysis highlights the value of leveraging BIM (Building Information Modeling) data to create a digital twin, as well as the improvements XR tools bring to the visualization of this data. In addition, a study of existing collaboration solutions revealed the requirements and constraints inherent in collaborative work between a field operation and a remote expert, such as the need for common work environment and difficulties in synchronizing exchanges. In a second phase of our research, based on these findings, we developed an architecture to enhance collaboration during inspections between an on-site operator and a remote expert, using the visualization capabilites of XR tools and data from the system’s digital twin. This led to the creation of the DT-RAMCoRE solution, enabling Information to be shared via indirect manipulation of the system’s digital twin usingthe RAMCoRE method. Aware of the diversity of XR devices and the rapid evolution of the technology, we designed this solution with the aim of ensuring its durability and interoperability between the different media available, thanks to the OpenXR standard. By analyzing a case study, we have demonstrated that using the DT-RAMCoREsolution during a collaborative inspection helps the operator to understand the information transmitted by an expert more quickly than with a conventional approach based on a video call, while reducing errors linked to the identification of system components. In conclusion, we formulate recommendations for optimizing the information provided to collaborative users. Finally, we have developed a simulation concept, a key component of the DT, to provide the field operator and the remote expert with advanced analysis tools to facilitate inspection work. As a perspective to this work, we present a methodology for exploiting DT simulation models via XR tools
Kherbache, Mehdi. „Toward Optimized 802.15.4 Industrial Wireless Networks : Harnessing Machine Learning and Digital Twins“. Electronic Thesis or Diss., Université de Lorraine, 2023. http://www.theses.fr/2023LORR0253.
Der volle Inhalt der QuelleThe Industrial Internet of Things (IIoT) presents a complex landscape with numerous constraints, particularly due to their use to control critical applications in Industry 4.0. The requirements in such a context in terms of energy efficiency and quality of service (delay, reliability, determinism and robustness) are strict and of paramount importance. Consequently, there is a pressing need for sophisticated management mechanisms throughout their entire lifecycle to meet these needs. This thesis explores two technological fronts to address this challenge: Reinforcement Learning-based Time Slotted Channel Hopping (TSCH) scheduling and Network Digital Twin (NDT). TSCH scheduling in IIoT, is identified as a crucial area to optimize the performance of these networks. Several works proposed Reinforcement Learning-based scheduling techniques for TDMA (Time Division Multiple Access ) MAC protocols, and particularly for TSCH. However, using this approach in a constrained network like the IIoT carries the risk of elevated energy consumption. This is due to the continuous learning process and coordination among the nodes necessary to manage the non-stationarity issue in the network, which is viewed as a Multi-Agent System. This thesis introduces a novel Reinforcement Learning-based distributed scheduling algorithm, QL-TSCH-plus. This algorithm has been designed to be adaptive and efficient, with reduced energy consumption and delay targets inherent to IIoT environments. Parallel to the development of TSCH scheduling, this thesis adopts the concept of NDT as a viable solution for effective IIoT management. Digital twins have been increasingly used to optimize the performances of industrial systems. Capitalizing on this technology, a holistic NDT architecture for the IIoT is proposed, where the network is integrated with other industrial components. The architecture leverages Software Defined Networking to enable closed-loop network management across the entire network life-cycle (from design to service). This architecture enables quick validation of networking solutions in an industrial environment because of the continuous link between the NDT and the physical IIoT network. Moreover, we propose to model the IIoT in the NDT using Petri-nets, enabling data-driven Petri-nets. These serve as coarse-grained formal models enabling fast simulation time for what-if scenarios execution, and real-time fault detection that is crucial in mission-critical industrial applications
Rifi, Léah. „Digital twin-based decision support system for the prospective and the retrospective analysis of an operating room under uncertainties“. Electronic Thesis or Diss., Ecole nationale des Mines d'Albi-Carmaux, 2023. http://www.theses.fr/2023EMAC0020.
Der volle Inhalt der QuelleWith healthcare demand rising worldwide, hospital services are increasingly needed. Hospitals' performance is tightly linked to their surgical suite performance. Indeed, the surgical suite is an important revenue and expense center with over 40% of the hospital's budget dedicated to it (Macario et al. 1997) and 60% of the patient coming into the hospital for surgical intervention (Fugener et al. 2017). This makes it necessary for surgical suites to be efficient. However, running a profitable surgical suite is quite hard and requires a methodological approach due to the complexity of its functioning: the diversity of patient pathways, the multiplicity of professions, the tight link with upstream and downstream wards, the synchronization of several resources and logistic flows (drug and medical devices), etc. On the other hand, durations variability and disruptions inherent in medical care like emergency cases are the main factors and events that degrade the scheduled execution and involve the staff making decisions frequently to preserve the surgical suite activity in an optimal way. Therefore, OR planning and scheduling activities are of increasing interest to the scientific community. In this PhD thesis, we focus on offline operational and online operational levels (Hans and Vanberkel 2012). This leads us to the following research questions: (1) How can we assess the robustness and the resilience of the schedule before its execution (prospective way)? (2) How can we replay the schedule to have feedback and assess the decisions made during its execution (retrospective way)? The contribution of this manuscript is threefold: (1) we propose a digital twin-based decision support system for the prospective and retrospective simulation and analysis of the operating room schedule execution, (2) we describe a standardized methodology to conceive, build and implement this tool in any surgical suite, (3) This methodology is applied to an operating room inspired by the Private Hospital of La Baie (Vivalto Santé group, France), in order to have a proof of concept allowing to simulate an operating program prospectively and retrospectively
Nerlikar, Vivek. „Digital Twin in Structural Health Monitoring for Aerospace using Machine Learning“. Electronic Thesis or Diss., université Paris-Saclay, 2023. http://www.theses.fr/2023UPASG080.
Der volle Inhalt der QuelleModern engineering systems and structures often utilize a combination of materials such as metals, concrete, and composites, carefully optimized to achieve superior performance in their designated functions while also minimizing overall economic costs. Primarily, engineering structures are subjected to dynamic loads during their operational life. The manufacturing issues and/or the perpetual dynamic operations often lead to some changes into a system that adversely impact its present and/or future performance; these changes can be defined as damage. The identification of damage is a crucial process that ensures the smooth functioning of equipment or structures throughout their life cycle. It alerts the maintenance department to take the necessary measures for repair. Structural Health Monitoring (SHM) is a potential damage identification technique which has attracted more attention in the last few decades. It has the capability to overcome the downsides of traditional Non-Destructive Testing (NDT). In this thesis, we used Ultrasonic Guided Waves (GW) technique for SHM. However, sensitivity of GW to Environmental and Operational Conditions (EOC) modify the response signals to mask defect signatures. This makes it difficult to isolate defect signatures using methods such as baseline comparison, where damage-free GW signals are compared with current acquisitions Baseline-free methods can be an alternative, but they are limited to simple geometries. Moreover, high sensitivity of GW to EOC and measurement noise poses a challenge in modelling GW through physics-based models. The recent advancements in Machine Learning (ML) has created a new modelling axis, including data-driven modelling and physics-based modelling, often referred to as Scientific ML. Data-driven modelling is extremely helpful to model the phenomena that cannot be explained by physics, allowing for the isolation of subtle defect signatures and the development of robust damage detection procedures. However, ML-based methods require more data to capture all the information to enhance the generalization capability of ML models. SHM, on the other hand, tends to generate mostly damage-free data, as damage episodes seldom occur. This particular gap can be filled through physics-based modeling. In this approach, the modeling capabilities of physics-based models are combined with measurement data to explain unexplainable phenomena using ML. The primary objective of this thesis is to develop a data-driven damage detection methodology for identifying defects in composite panels. This methodology is designed for monitoring similar structures, such as wind or jet turbine blades, without requiring pristine (damage-free) states of all structures, thereby avoiding the need for direct baseline comparisons. The second goal is to develop a physics-based ML model for integrating physics-based simulations with experimental data within the context of a Digital Twin. The development of this physics-based ML model involves multi-fidelity modeling and surrogate modeling. To validate this model, we utilize an experimental and simulation dataset of an Aluminium panel. Furthermore, the developed model is employed to generate realistic GW responses at the required damage size and sensor path. These generated signals are then used to compute a Probability of Detection (POD) curve, assessing the reliability of a GW-based SHM system
Hosni, Houssem. „Conception d’un jumeau numérique pour un procédé d’aspiration industrielle“. Electronic Thesis or Diss., La Rochelle, 2022. http://www.theses.fr/2022LAROS031.
Der volle Inhalt der QuelleThis thesis is devoted to the design of a digital twin for industrial ventilation systems in a monitoring and diagnostic context. The input measurements of the model are exclusively electrical and no mechanical sensors are used. The particular cases of fixed and variable speed are studied. At constant speed, the demodulation of electrical currents is particularly studied and an original algorithm, based on an orthogonal signal generator, is proposed, compared to the main known methods, and validated experimentally. At variable speed, the proposed approach is based on order tracking methods without mechanical sensor in which the analysis signals are sampled as a function of the mechanical angle. In this case, the spectral components become independent of the rotation speed and the frequency analysis can be exploited. An original method is presented. It is based on the definition of an observer from a reduced number of information on the considered motor. The estimated speed is used to deduce the mechanical position and to perform the angular resampling. A generalization of the concept of ordertracking is also presented, allowing to define resampling angles adapted to the monitored faults. This method is tested on a LIAS test bench and on the industrial ventilation process
Tchana, De Tchana Yvan. „Proposition d’un jumeau numérique pour soutenir la gestion de l'exploitation d'une infrastructure linéaire“. Thesis, Troyes, 2021. http://www.theses.fr/2021TROY0012.
Der volle Inhalt der QuelleThe digital growth of the construction industry led to BIM (Building Information Modeling). Developed for buildings, BIM is later used on linear infrastructure projects. Such projects require end-to-end control of information. PLM (Product Lifecycle Management) supports digital continuity in the manufacturing industry. Studies evaluate the relevance of a complementary use of the BIM and PLM approaches for linear infrastructure projects. With an adaptation of methods used for building construction, those studies are mostly restricted to the implementation of data repositories. This makes it difficult to consider the infrastructure post-construction phase, where the 3D model is no longer a digital model, but a digital twin. This research work consists in developing a strategy for the design, the implementation and the operations and maintenance of a linear infrastructure. The digital twin of the infrastructure is the target of our approach. It will take into consideration not only BIM and PLM methodologies, but also any other data source positioning the infrastructure in its geographical environment. Data aggregator, our digital twin should make it possible to manage the lifecycle of a linear infrastructure. This system is tested on a specific linear infrastructure, a level crossing. Digital continuity and data traceability are important factors for those constructions. Through the digital twin, our proposal helps to follow the data, and thus to link operational data to the design and construction data of the linear infrastructure
Bouzid, Sami. „Jumeau numérique temps réel de machines électriques basé sur la méthode des éléments finis“. Master's thesis, Université Laval, 2020. http://hdl.handle.net/20.500.11794/40132.
Der volle Inhalt der QuelleVallet, Yves. „Contribution à la caractérisation et à la modélisation de l'accouchement instrumenté par ventouse“. Electronic Thesis or Diss., Université de Lorraine, 2023. http://www.theses.fr/2023LORR0134.
Der volle Inhalt der QuelleDuring vaginal delivery, the expulsion phase is associated with risks of maternal and fetal complication. Practitioners may need to perform a vacuum assisted delivery (VAD). As with any operational instrument, there are risks associated with its use, for the baby and the mother, if it is not carried out correctly. For the fetus, the various layers of the scalp are solicited with great strain, which can lead to rare complications such as caput succedaneum, cephalohaematomas and subgaleal haemorrhage. To limit these risks, practitioners must be trained to use suction cups, and must comply with current recommendations. However, the parameters associated with its use, such as the amplitude of the traction force, the placement of the suction cup and the extraction procedure, remain operator-dependent. A better understanding of the physical and mechanical parameters involved in VAD is therefore needed to improve this practice. In order to achieve this objective, this thesis is structured around two axes, corresponding to two different scales and including experimental and numerical work. The first, at the macroscopic scale, considers the fetus in its environment. The second, at the mesoscopic scale, considers the isolated fetal head within its environment. The work of the first part has allowed to capture gesture of the the practitioners on a training dummy and to create a digital twin of this didactic tool in order to study the various parameters of VAD. In the second part, the work has led to the development of a model of the interaction between the scalp's skin and skull. A numerical model was then designed and implemented using a refined" modelling of the fetal head to study the parameters of the VAD. In parallel, a review of the literature on suction cups found in nature open up promising prospects for the evolution of the obstetric suction cup design used nowadays
Lehmann, Fanny. „A surrogate model of elastic wave propagation to quantify uncertainties in seismic hazard analysis“. Electronic Thesis or Diss., université Paris-Saclay, 2024. http://www.theses.fr/2024UPAST074.
Der volle Inhalt der QuelleThe propagation of seismic waves in the ground is subject to many sources of uncertainties, ranging from the uncertain activity of geological faults to the incomplete knowledge of mechanical properties inside the Earth's crust. To properly assess seismic hazard, it then becomes essential to quantify how uncertainties influence the intensity of ground motion generated by earthquakes.In areas with low-to-moderate seismicity, like most regions in metropolitan France, seismic records are too sparse to evaluate ground motion uncertainties. In this situation, numerical simulations are the only option to estimate ground motion intensity, but their high computational costs prevent most uncertainty analyses. In this thesis, we design a surrogate model that can replace the numerical solver by drastically reducing the computational costs while preserving its flexibility and a satisfying accuracy.We first illustrate the influence of geological heterogeneities on ground motion intensity in the context of the Mw4.9 Le Teil earthquake (Ardèche, France, 2019). Heterogeneities are added to a regional geological model in the form of random fields, and we show that it generates more realistic ground motion. However, heterogeneities also lead to a large variability between samples.To study this variability systematically, we build a database of 30,000 heterogeneous 3D geological models, and inside each geology, seismic waves are propagated from a random source using the spectral element code SEM3D. The database is then used to train a surrogate model in a purely data-driven framework.To design the surrogate model, we propose an extension of the Fourier Neural Operator called the Multiple Input Fourier Neural Operator (MIFNO). The MIFNO takes as inputs a 3D geology and a vector of source parameters to predict 3D ground motion. Ground motion is a time-dependent surface wavefield, but we do not need any time iteration thanks to a depth-to-time conversion. We characterize the MIFNO prediction error and explore the MIFNO generalization ability to out-of-distribution data.We finally take advantage of transfer learning to further improve the MIFNO accuracy in the context of the Le Teil earthquake. With this fine-tuned surrogate model, we obtain statistical distributions of several quantities of interest in seismic hazard assessment. They are coherent with numerical simulations and provide confidence intervals that were out of reach with existing methods
Gregorio, Jean-Loup. „Contribution à la définition d'un jumeau numérique pour la maîtrise de la qualité géométrique des structures aéronautiques lors de leurs processus d'assemblage“. Thesis, université Paris-Saclay, 2020. http://www.theses.fr/2020UPASN008.
Der volle Inhalt der QuelleAssembly operations of aerostructures are nowadays planned using the Digital Mock Up of the considered products. To allow a good realization of the aforementioned operations, the geometry of the physical product must remain as faithful as possible to the reference geometry contained in the Digital Mock Up. Because of the complexity of the considered products, unanticipated geometrical deviations may however appear. These geometrical deviations lead to longer cycle times and higher assembly costs.The increasing integration of data processing system gives new prospects on how to organize production systems. These prospects include the possibility to optimize the manufacturing and assembly operations in real time thanks to the use of digital twins of the manufactured products. In this work, we propose the implementation of a geometrical digital twin. This digital twin is capable of mirroring the geometry of the physical product being assembled and optimizing the geometry of some components remaining to be assembled.With this in mind, the initial Digital Mock Up is updated in order to obtain a hybrid representation of the product. This representation includes the different states of the components, which are called as-designed, as-built and interface. The as-built components are more particularly updated in order to mirror the geometry of the physical product being assembled. A method using digitized data is proposed. From there, the geometry of interface components is updated so that the final product complies with the functional requirements which were defined. A method is also proposed for this purpose.The feasibility of the approach as well as the proposed tools is evaluated through two application cases, one of which is directly representative of the industrial context of the works. The obtained results allow to consider enriching the proposed approach by considering non-geometrical constraints in order to optimize assembly operations
Chaufer, Martin. „Développement d’un substitut physique de thorax humain et de son jumeau numérique dédiés à la prédiction du risque lésionnel lors d'impacts balistiques non pénétrants“. Electronic Thesis or Diss., Bourgogne Franche-Comté, 2023. http://www.theses.fr/2023UBFCA015.
Der volle Inhalt der QuelleIn recent years, the use of less-lethal weapons has increased. These weapons, designed to neutralise individuals exhibiting dangerous behaviour, can cause injuries or even death. Similar injury mechanisms are observed in the rear deformation of bulletproof vests during impacts. To protect citizens and law enforcement personnel, it is necessary to prevent such scenarios. However, today there are few tools available to assist in the sizing of such equipment. In this context, this thesis aims to develop tools for predicting thoracic injury risk during non-penetrating ballistic impacts. Accordingly, a physical substitute of the human thorax and its numerical twin are developed. Initially, the HUByx numerical model is used as a reference to construct a simplified numerical model that can be manufactured using readily available materials. Different materials are characterised, and their material laws are established. Once validated, this numerical model serves as a basis for constructing the physical substitute called SurHUByx. It is equipped with various sensors to record data over the rib and in internal organs during ballistic impacts. Specific impact cases described in case reports are replicated on SurHUByx to correlate sensor data with injury assessments. Finally, a statistical approach is used to develop injury prediction curves, allowing to estimate of the risk of injury following an impact on SurHUByx or its numerical twin, SurHUByx FEM
Levy, Benjamin. „Étude numérique et expérimentale pour le développement d’un nouveau procédé de tribo-grenaillage“. Thesis, Paris, HESAM, 2021. http://www.theses.fr/2021HESAE018.
Der volle Inhalt der QuelleShot peening is a surface treatment commonly applied in the aerospace, automotive and biomedical industries to improve the mechanical performance of parts. This treatment consists in introducing residual compressive stresses in the sub-surface. However, technological advances, accompanied by the evolution of materials, have generated new demands in terms of shot peening treatment. In particular, the industrial need for a treatment capable of both ensuring a sufficient level of mechanical performance while functionalizing the surface is increasingly felt. The aim of this work is to show to what extent this need can be met by a new treatment called tribo-peening. The two functions targeted by tribo-peening require a characterization of the surface states (functionalization) and sub-surface (residual stresses) of the treated parts. These states are the result of mechanical interactions between media, of different nature and shape, and the treated surface. The tribo-peening approach consists of controlling these interactions, the texture and energy transfers involved in order to master the implanted functional signature. Therefore, a multi-scale characterization of the target surface and sub-surface is performed simultaneously with that of the media surface. This characterization step is based on the evaluation of tribo-peened surface representative of the overall texturing, the so-called Elementary Representative Areal Surface. The control and optimization of the process are envisaged through the establishment of a digital twin fed with multi-scale characterization data, finite element modeling as well as data from the instrumentation of the physical twin
Ledezma, Lopez Gabriel Alejandro. „Suitable representations of gamma alumina porous structures by computational modeling“. Thesis, Lyon, 2021. https://tel.archives-ouvertes.fr/tel-03789640.
Der volle Inhalt der QuellePorous materials are widely used in chemical engineering. At the mesoporous scale, confinement effects influence the thermodynamic of the system and the transport conditions. Indeed, the architecture of the pore network is the origin of mass transfer limitations within disordered porous materials. Therefore, it is important to understand not just the textural properties of the solid but also its pore network characteristics. Gamma-alumina is a disordered porous material with an elevated tortuosity very often used in the oil refining and petrochemistry, whose topology is not yet fully understood (1, 43, 44). Recent research articles propose that this material has different pore domains, each one characterized by its own pore size distribution and void fraction (45). The interplay among these different levels clearly plays a role in effective diffusion. This work intends to better understand the textural and topological descriptors of gamma alumina by creating a digital representation of it. The catalyst is represented using a pore network model. The pore network representation is then characterized using originally developed computational equivalents of textural and mass transfer characterization techniques. The experimental validation was done through the generation of digital twins for real gamma alumina samples. Using diffusion simulations on the pore network model fitted to the nitrogen sorption curves, a tortuosity factor was predicted that differs by less than 20% from the tortuosity factor measured by PFG-NMR. This illustrates how a digital twin allows to provide a reasonable estimate for the tortuosity factor from readily available nitrogen porosity experiments. The research work in this thesis is the start of the path to ultimate goal of improving the catalytic performance of disordered porous catalyst by the digital optimal design of the material architecture. At the same time, the accuracy of the models used to design and evaluate heterogeneous reactor performance will be improved
Somera, Audrey. „On the effective elasticity of quasi-periodic lattice materials : From microscopic foundations to experimental validation“. Electronic Thesis or Diss., Ecole centrale de Nantes, 2022. http://www.theses.fr/2022ECDN0037.
Der volle Inhalt der QuelleArchitectured materials have received increasing interest over the years, especially by allowing new areas of the Ashby diagrams to bereached. Quasi-periodic lattices combine the advantages of both random and periodic structures: they are deterministic structures, their behaviour is isotropic, and they have better toughness than periodic lattices. However, the study of the mechanical behaviour of such structures is still in its infancy. Thus, this thesis proposes to study the effective elastic behaviour of quasi-periodic lattices. First, the local deformation mechanisms of different patterns have been studied. It is shown that the patterns could be separated into three categories: the completely stretching and bendingdominated patterns and the variable dominance ones. The influence of these local mechanisms on the overall mechanical behaviour was then investigated. For this purpose, an identification procedure of the lattice equivalent effective behaviour, based on a FEMU-type method, was implemented. First performed using a numerical twin, an experimental set-up was then designed to carry out the procedure and validate the numerical results experimentally. It is shown that the most suitable behaviour model depends on the pattern considered. While a classical Cauchy-type law seems sufficient to describe the behaviour of completely stretching-dominated and variable dominance patterns, it is necessary to use a Cosserat-type model for completely bending-dominated ones
Guillot, Matthieu. „Le problème du plus court chemin stochastique et ses variantes : fondements et applications à l'optimisation de stratégie dans le sport“. Thesis, Université Grenoble Alpes, 2020. http://www.theses.fr/2020GRALM024.
Der volle Inhalt der QuelleA golf course consists of eighteen holes. On each hole, the golfer has to move the ball from the tee to the flag in a minimum number of shots. Under some assumptions, the golfer's problem can be modeled as a stochastic shortest path problem (SSP). SSP problem is a special case of Markov Decision Processes in which an agent evolves dynamically in a finite set of states. In each state, the agent chooses an action that leads him to another state following a known probability distribution. This action induces a cost. There exists a `sink node' in which the agent, once in it, stays with probability one and a cost zero. The goal of the agent is to reach the sink node with a minimum expected cost. In the first chapter, we study the SSP problem theoretically. We define a new framework in which the assumptions needed for the existence of an optimal policy are weakened. We prove that the most famous algorithm still converge in this setting. We also define a new algorithm to solve exactly the problem based on the primal-dual algorithm. In the second chapter we detail the golfer's problem model as a SSP. Thanks to the Shotlink database, we create `numerical clones' of players and simulate theses clones on different golf course in order to predict professional golfer's scores. We apply our model on two competitions: the master of Augusta in 2017 and the Ryder Cup in 2018. In the third chapter, we study the 2-player natural extension of SSP problem: the stochastic shortest path games. We study two special cases, and in particular linear programming formulation of these games
Garcia, David. „Études exploratoires dédiées au diagnostic de corrosion assisté par ordinateur des structures de génie civil“. Thesis, Toulouse 3, 2020. http://www.theses.fr/2020TOU30247.
Der volle Inhalt der QuelleThe PhD thesis "Exploratory studies dedicated to computer-assisted corrosion diagnosis of civil engineering structures" deals with the phenomenology and modeling of corrosion of structural steel. The safety, societal and environmental impact of aging infrastructures makes this theme a major economic issue for the development of any country. The proposed developments focus mainly on the corrosion of reinforcements in reinforced concrete. The corrosion of buried metallic structures is also addressed concerning the problems related to galvanic couplings induced by the heterogeneity of soils and stray currents. The usual methods of investigation (measurements of steel potential, concrete resistivity or polarization resistance), combined with empirical hypotheses established by experience, lead to interpretations that are often uncertain or have only a qualitative value. The ambition of this thesis, motivated by the issues at stake, is to show how a better understanding of the physics of corrosion, combined with the power of finite element calculation, allows the construction of elaborate and robust models, useful for a quantified and reliable diagnosis and/or prognosis. The thesis is abundantly illustrated by real or numerical case studies and supported by original laboratory tests. In order to improve the understanding of the phenomena prevailing in the corrosion process, the key concepts of thermodynamics and electrochemical kinetics are recalled and contextualized. The assembly of different physical, chemical and electrochemical laws allows the elaboration of an advanced modeling approach, integrating in particular the diffusion of oxygen to the reinforcement in an unsaturated context, but also the production and precipitation of corrosion products and their influence on the dynamic equilibrium of a corrosion system. This modeling approach, necessarily three-dimensional or at least two-dimensional, gives rise to a transcription in a finite element calculation code. It is first applied to the numerical study of a first typical case of corrosion: a reinforced concrete pile partially submerged in the sea. The influence of the role of oxygen (availability and diffusion) on the dissolution kinetics of the steel and on the nature of the corrosion products formed is studied in particular. In order to illustrate the effective contribution of 3D modeling in the process of corrosion diagnosis, a real case study is proposed concerning a buried steel structure, in this case sheet piles used to support the abutments of a freeway overpass, located near a pipe buried under cathodic protection. Measurements carried out in-situ but also in the laboratory from judiciously chosen samples are used to feed the calculation model. The numerical model thus constructed, qualified as a digital twin, makes it possible to highlight the existence of stray currents circulating in the structure, but also the risk of galvanic corrosion induced by the heterogeneity of the soil. The electrochemical digital twin is then a powerful tool for estimating the kinetics and the corrosion facies of the structure and making a prognosis in terms of service life. Within a concrete structure, the presence of chlorides is associated with various effects, notably associated with the local electric field. If this phenomenon is ignored, the interpretation of field data, for example potential maps, can lead to a biased diagnosis. This thesis addresses the question of corrosion initiation.[...]
Tahiri, Imane. „Contribution à la conception d'une commande reconfigurable et tolérante aux fautes pour les Systèmes Automatisés de Production“. Thesis, Reims, 2020. http://www.theses.fr/2020REIMS008.
Der volle Inhalt der QuelleThe research results presented in this thesis falls within the framework of control reconfiguration of Automated Production Systems (APS) seen as a class of Discrete Event Systems (DES). The reconfiguration is based on Supervisory control theory (SCT) and triggered following a plant fault detection (fault tolerant control). The faults considered in this report are sensor faults (a sensor which remains blocked on reading 1 or on reading 0). The main contribution is based on a safe control synthesis founded on timed properties. In fact, if a sensor fault is detected, the controller of the normal behavior is reconfigured to a timed one where the timed information compensates the information lost on the faulty sensor. The switch between normal and reconfigured behaviors is ensured using reconfiguration rules. The main objective of our method is to implement the obtained control into a PLC. To meet the different objectives, we propose a method to translate the distributed controllers of the two operating modes, as well as the reconfiguration rules into different Grafcets implantable in a PLC. The implementation of these different models is verified by a model-checking technique before being tested on a digital twin. Finally, we apply our contribution to a real system to illustrate our results
Ranjbar, Gigasari Roza. „Model Predictive Controller for large-scale systems - Application to water networks“. Electronic Thesis or Diss., Ecole nationale supérieure Mines-Télécom Lille Douai, 2024. http://www.theses.fr/2024MTLD0002.
Der volle Inhalt der QuelleThis thesis addresses the challenge of optimizing the management of canals, a complex task due to their extensive scale and distinctive attributes, including intricate dynamics, considerable time delays, and minimal bottom slopes. Specifically, the central goal is to ensure the navigability of the network, which involves maintaining safe water levels for vessel travel, through control theory. More precisely, the water levels must remain within a predefined range around a setpoint. Additionally, typical aims encompass reducing operational costs and enhancing the equipment’s life expectancy. In this regard, another objective in the management of such networks is replacing the possible sensors across canals by applying a moving robot to take the required measurements. To accomplish effective management, it becomes imperative to ensure efficient control over hydraulic structures such as gates, pumps, and locks. To this end, a control algorithm is introduced based on an existing model derived from the Saint-Venant equations. The modeling approach simplified the original complex description providing adaptability and facilitating the systematic integration of both current and delayed information. However, the resulting model formulation falls within the category of delayed descriptor systems, necessitating extensions to standard control and state estimation tools. Model predictive control and moving horizon estimation methods can be readily tailored for this formulation, while also adapting physical and operational constraints seamlessly. Given the extensive nature of canals, an evaluation of the digital twin was untaken to address the critical need for advanced tools in the management of such networks. By harnessing the capabilities of digital twins, we aimed to enhance our understanding of canal dynamics, past scenarios, and management strategies. This evaluation sought to bridge the gap between theory and practical implementation, offering a tangible means to playback past events, test diverse management approaches, and ultimately equip decision-makers with robust criteria for informed and effective network management.The methodologies presented above are applied to a practical case study, a canal in the northern region of France. The objective is to validate the efficacy of these approaches in a real-world context.While centralized MPC provides resilience through its receding-horizon approach, its deterministic nature limits its ability to systematically address uncertainties. To effectively tackle these system uncertainties, the implementation of Stochastic MPC (SMPC) has been adopted. SMPC integrates probabilistic descriptions into control design, offering a methodical approach to accommodating uncertainties. In this context, the application of SMPC is interconnected with a mobile robot aimed at replacing existing sensors along the canal to capture measurements. Consequently, a part of this thesis focuses on the design of SMPC in conjunction with a mobile robot. This approach has been applied to an ASCE Test canal to evaluate its effectiveness