HAL-MINES ParisTech
Not a member yet
27348 research outputs found
Sort by
A deep learning approach for time-consistent cell cycle phase prediction from microscopy data
No full textInternational audienceThe cell cycle is a series of regulated stages during which a cell grows, replicates its DNA, and divides. It consists of four phases -two growth phases (G1 and G2), a replication phase (S), and a division phase (M) -each characterized by distinct transcriptional programs and impacting most other cellular processes. In imaging assays, the cell cycle phase can be identified using specific cell-cycle markers. However, the use of dedicated cell-cycle markers can be impractical or even prohibitive, as they occupy fluorescent channels that may be needed for other reporters. To address this limitation we propose a method to infer the cell cycle phase from a widely used fluorescent reporter: SiR-DNA, thereby bypassing the need for phase-specific markers while leveraging information already present in common experimental setups.Our method is based on a Variational Auto-Encoder (VAE), enhanced with two auxiliary tasks: predicting the average intensity of phase-specific markers and enforcing temporal consistency through latent space regularization. The reconstruction task ensures that the latent space captures cell cycle-relevant features, while the temporal constraint promotes biological plausibility. The resulting model, CC-VAE, classifies cell cycle phases with high accuracy from widely used DNA markers and can thus be applied to high-content screening datasets not specifically designed for cell cycle analysis. CC-VAE is freely available, along with a new, publicly released dataset comprising over 600,000 labeled HeLa Kyoto nuclear images to support further development and benchmarking in the community.</div
Evaluating fault detection techniques for refrigerant leak detection in industrial vapor compression refrigeration systems
No full textInternational audienceEfficient detection of refrigerant leakage is essential for industrial refrigeration systems due to its considerable impact on system performance and the environment. Refrigerant leaks can be detected by monitoring the drop in the liquid level within the receiver.In this work, we evaluate fault detection techniques for identifying refrigerant leaks, comparing a fixed threshold method, such as Exponentially Weighted Moving Average, with a recent dynamic threshold approach. We also propose improving the latter method, achieving superior performance in balancing the trade-off between detecting genuine leaks and minimizing false alerts
Simulation de propagation d'ondes ultrasonores dans les matériaux polycristallins
No full textThe metallurgical industry plays a crucial role in the industrial sector, particularly in high-stakes fields such as nuclear, aerospace and defense engineering. Ultrasonic Testing (UT) is widely employed to ensure the quality of forged metallic components. However, practitioners face significant challenges, notably due to signal pollution caused by spurious reflections of ultrasonic waves from the polycrystalline microstructure of metals. In this context, a high-fidelity numerical simulation tool would enable a more accurate characterization of ultrasonic waves behavior as they propagate through polycrystalline materials, allowing a better understanding of the received signals. Furthermore, advanced techniques such as laser-based Ultrasonic Testing for monitoring thermomechanical processes are currently under development. The coupling of numerical simulation tools for microstructure evolution during material forming with those dedicated to ultrasonic wave propagation offers a promising approach. This PhD work was dedicated to the development of a high-fidelity simulation tool, building upon the state of the art and following three main steps:1. defining a model to describe ultrasonic wave propagation in polycrystals,2. developing an appropriate numerical scheme to solve this model,3. validating the resulting model.To achieve this, the developed model is based on the first-order velocity-stress ElastoDynamics wave propagation equations. This model is solved using a Discontinuous Galerkin spatial scheme, combined with explicit time integrators such as Leap-Frog and Runge-Kutta methods. Finally, the features and performance of the resulting simulation environment are assessed through academic test cases before being applied to numerical polycrystalline microstructures, closely resembling configurations encountered in experimental Ultrasonic Testing.L'industrie métallurgique occupe une place importante dans de nombreux secteurs stratégiques, en particulier dans des domaines de haute exigence tels que le nucléaire, l'aéronautique et la défense. Le contrôle par ultrasons y est couramment utilisé afin de garantir la qualité des composants métalliques forgés. Toutefois, les praticiens sont souvent confrontés à des difficultés, notamment en raison de la pollution du signal causée par les réflexions parasites des ondes ultrasonores sur la microstructure polycristalline des matériaux métalliques. Dans ce contexte, un outil de simulationnumérique haute-fidélité permettrait de caractériser avec précision le comportement des ondes ultrasonores lors de leur propagation dans les polycristaux, et de mieux interpréter les signaux reçus.Par ailleurs, des techniques avancées, telles que le contrôle ultrasonore par laser pour le suivi in situ des microstructures lors de procédés à chaud, sont en cours de développement. Le couplage entre les outils numériques de simulation des évolutions de microstructures à l'œuvre lors de procédés de mise en forme et ceux dédiés à la propagation des ondes ultrasonores constitue une approche prometteuse. C'est dans cette optique que ce travail s'est inscrit, pour développer un outil de simulation haute-fidélité polyvalent en nous appuyant sur l'état de l'art et en suivant trois étapes :1. la définition d'un modèle représentant la propagation des ondes ultrasonores dans les matériaux polycristallins éventuellement multiphasés,2. le développement d'un schéma numérique adapté pour résoudre ce modèle,3. et la validation du modèle obtenu.À cette fin, nous nous sommes appuyés sur les équations de propagation des ondes issues de l'élastodynamique d'ordre 1 en vitesse-contrainte. Ce modèle est résolu à l'aide d'un schéma numérique de type Galerkin discontinu en espace, couplé à des intégrateurs temporels explicites de type Saute-Mouton et Runge-Kutta. Enfin, les performances et les caractéristiques de l'environnement de simulation ainsi conçu ont été évaluées à travers des cas académiques, avant d'être appliquées à des microstructures polycristallines numériques, reproduisant des configurations proches de celles rencontrées en contrôle ultrasonore expérimental
A Real-Time Rig Control ROP Optimization Framework Using Machine Learning and Predictive Vibration Modeling
No full textInternational audienceAbstract This paper presents a novel real-time rig control framework for optimizing the rate of penetration (ROP) in horizontal drilling operations by integrating machine learning (ML), transfer learning, and predictive vibration modeling within a constraint-aware optimization loop. A general ROP prediction model is first trained on a large, multi-well dataset using ensemble learning methods, with XGBoost emerging as the best-performing algorithm. To improve prediction accuracy in new wells, a transfer learning strategy fine-tunes the model using real-time and offset well data, ensuring adaptability to local geological conditions. The fine-tuned model is embedded within an optimization algorithm that recommends drilling parameters to maximize ROP while respecting operational constraints. These include rotary speed limits derived from a predictive vibration model that identifies resonance zones to prevent harmful drilling vibrations. Additional constraints ensure parameter feasibility, operational continuity, and adherence to target depth of cut specifications. The framework is validated in both simulated and real-world environments. In simulation, a physics-based drilling model is used to generate ground-truth data for assessing convergence and performance. Results show that the optimization algorithm reliably steers drilling parameters toward the region of maximum ROP. Field validation on real rig data further confirms the robustness and practicality of the proposed approach, demonstrating significant improvements in ROP prediction accuracy and drilling performance. This hybrid data-physics optimization framework enables safer, more efficient drilling by providing actionable, real-time parameter recommendations that account for both data-driven insights and physical system dynamics
Integrated Multiphysics Modeling of a Ship’s Propulsion, Thermal, and Electrical Systems in Matlab/Simulink
No full textInternational audienc
Hygrothermal performance of wood-cement walls across various climate conditions
No full textInternational audienc
Multi-objective hyperparameter optimization of artificial neural network in emulating building energy simulation
No full textInternational audienc
Intégration de la vulnérabilité aux canicules dans la rénovation des maisons
No full textIn a context of global warming, the risks associated with heatwaves in housing are becoming a major issue. While these risks are now considered in the construction of new housing, they are often overlooked in existing buildings. In fact, renovation today is primarily focused on reducing heating consumption, which results, on the one hand, in increased health risks during heatwaves and, on the other hand, in the widespread adoption of inefficient air conditioning systems in homes. This thesis aims to provide building professionals with a decision-support tool to integrate the issue of heatwave vulnerability into house renovation. The tool must meet the constraints of renovation, which mean that dynamic energy simulation is rarely performed — in other words, it must be easy to implement and configurable using available data. It should allow for the evaluation of the impact of passive solutions on heatwave resistance and help determine whether installing an air conditioning system is necessary. To meet this objective, a methodology was developed based on the use of dynamic energy simulation withparticular attention paid to the modeling of natural ventilation and the ground. In a first phase, heatwave sequences are built and comfort and risk indicators are proposed, followed by a parametric study conducted on two common types of singlefamilyhouses in France. In a second phase, a simplified heatwave vulnerability indicator is developed. The simulation results make it possible to evaluate the impact of energy renovation scenarios depending on the two contrasting house types and their locations, as well as to assess the potential of passive overheating reduction strategies. Finally, the heatwave resistance indicator supports project owners in analyzing the need for adaptation and choosing appropriate measures for occupants.Dans un contexte de réchauffement climatique, les risques associés aux canicules dans les logements deviennent un problème majeur. Si ces risques sont désormais considérés dans la construction de nouveaux logements, ils sont souvent négligés dans l’existant. En effet, la rénovation est aujourd’hui essentiellement axée sur la réduction de consommations de chauffage avec pour conséquence d’une part une augmentation des risques sanitaires lors des vagues de chaleur et d’autre part, une diffusion massive d’une climatisation peu performante dans les logements. Cette thèse vise à proposer aux acteurs du bâtiment un outil d’aide à la décision pour intégrer la question de la vulnérabilité aux canicules dans la rénovation des maisons. L’outil doit répondre aux contraintes en rénovation qui font que la simulation énergétique dynamique est rarementréalisée, c’est-à-dire, être facile à mettre en œuvre et paramétrable à partir des données disponibles. Il doit permettre d’évaluer l’impact de solutions passives sur la résistance aux canicules et de trancher sur le besoin d’installer un système de climatisation, le cas échéant. Pour répondre à cet objectif, une méthodologie a été mise en place, basée sur l’utilisation de la simulation énergétique dynamique avec une attention particulière accordée à la modélisation de la ventilation naturelle et du sol. Dans une première phase, des séquences caniculaires sont construites et des indicateurs de confort et de risque sont proposés, puis une étude paramétrique est menée sur deux types de maisons individuelles courantes en France. Dans une deuxième phase, un indicateur simplifié de vulnérabilité aux canicules est développé. Les résultats de simulation permettentd’évaluer l’impact de scénarios de rénovation énergétique en fonction des deux types contrastés de maison et de leur localisation ainsi de que d’évaluer le potentiel de stratégies passives de réduction de la surchauffe. Enfin, l’indicateur de résistance aux canicules permet d’accompagner une maîtrise d’ouvrage dans l’analyse du besoin d’adaptation et le choix de mesures appropriées pour les occupants
Distributed Economic Dispatch in Power Networks Incorporating Data Center Flexibility
No full textInternational audienceWe consider Data Centers (DCs) as flexible loads that can alter their power consumption to alleviate congestion in the electric power network. We model DCs using a queuing-theoretic view and we form a Quality of Service (QoS)-based cost function that signifies how well a DC can carry out its workload given an amount of active servers. We integrate DCs in a centralized economic dispatch problem that determines, apart from power generation, DC workload shifting and server utilization, while respecting transmission line constraints. We further present a tractable decentralized formulation obtained via Lagrangian decomposition, which we solve using a dual gradient ascent algorithm. Experimental results on a standard power network explore the system-wide benefits of DC flexibility in "coupled" data and power networks, emphasizing on the trade-offs between the DC location, QoS, and efficiency.</div
