Université de technologie de Troyes open archive
Not a member yet
10722 research outputs found
Sort by
Solution methods for the VRP with sharing of deliveries between producers in local food logistic
No full textInternational audienceThis study addresses the logistical challenges in Short Food Supply Chains (SFSC), where producers face high logistical costs. A new variant of the Vehicle Routing Problem (VRP) is proposed, enabling producers to share their deliveries with partner producers. With a partner producer, a producer can either drop off goods for a customer (to be served by the partner producer) or pick up goods for a customer they will visit themselves during their routes.Although solving the Vehicle Routing Problem (VRP) is a key area in optimization, with certain variants tailored to the specificities of SFSC, such as Pickup and Delivery or split deliveries, few studies address a delivery sharing aspect like this.Three heuristic methods were tested: GRASP (Greedy Randomized Adaptive Search Procedure), ILS (Iterated Local Search), and VNS (Variable Neighborhood Search), which systematically explores multiple neighborhoods. Initial results show a significant reduction in logistical costs thanks to the sharing approach, while also highlighting an imbalance in the workload distribution between producers.Mots clés -Circuits CourtsAlimentaires de Proximité, VRP, mutualisation, méta-heuristiques Résumé Cette étude traite des défis logistiques dans les Circuits Courts Alimentaires de Proximité (CCAP), où les producteurs sont confrontés à des coûts logistiques élevés. Une nouvelle variante du problème de tournées de véhicules (VRP) est proposée, permettant aux producteurs de mutualiser leurs livraisons avec d'autres producteurs partenaires. Avec un producteur partenaire, un producteur peut soit lui déposer des marchandises pour un client (que ce producteur servira), soit prendre des marchandises pour un client qu'il visitera lui-même lors de ses tournées. Bien que la résolution du Problème de Tournées de Véhicules (VRP) soit un domaine clé en optimisation, avec certaines variantes adaptées aux spécificités des CCAP, comme le Pickup and Delivery ou les livraisons fractionnées, peu d'études abordent un aspect de mutualisation des tournées comme celui-ci. Trois méthodes de résolutions approchées ont été testées : GRASP (Greedy Randomized Adaptative Search Procedure), ILS (Iterated Local Search), et VNS (Variable Neighborhood Search), qui explore plusieurs voisinages de manière systématique. Les premiers résultats montrent une réduction significative des coûts logistiques grâce à la mutualisation, et mettent en évidence un déséquilibre dans la répartition des charges entre producteurs.</div
Linear regression for currency European call option pricing in incomplete markets
No full textInternational audienceThe least squares is the traditional regression technique for pricing European options in incomplete markets by building a self-financing hedging portfolio that does not perfectly replicate the call option.However, the least squares is quite sensitive to even a single outlier in the data, and thus the predicted option price may potentially deviate from the true unknown one. To alleviate the problem of outliers, this chapter aims to develop two different option pricing prediction strategies based mainly on the idea of robust linear regression. The robust techniques proposed are evaluated on numerical data, the results of which demonstrate their effectiveness for European call option pricing on exchange rates
Optimizing Recycling Processes for Mixed LFP/NMC Lithium-Ion Batteries: A Comparative Study of Acid-Excess and Acid-Deficient Leaching
No full textInternational audienceThis study explores the optimization of hydrometallurgical processes for recycling lithium-ion batteries (LIBs) containing a mixture of lithium iron phosphate (LFP) and nickel-manganese-cobalt (NMC) cathodes. Two approaches were investigated: acid-excess leaching and acid-deficient leaching with residue recirculation. A design of experiments (DoE) framework was applied to assess the impact of key parameters, including sulfuric acid and hydrogen peroxide concentrations, as well as solid-to-liquid (S/L) ratios, on the dissolution yields of target metals (Ni, Mn, Co, and Li). Acid-excess leaching achieved nearly complete dissolution of target metals but required additional purification steps to remove impurities. Acid-deficient leaching with a 60% recirculation of leaching residue improved dissolution yields by up to 12.5%, reduced reagent consumption, and minimized operational complexity. The study also evaluated separation strategies for manganese and cobalt through solvent extraction. Results indicate that while acid-excess leaching offers higher yields, acid-deficient leaching with residue recirculation is more cost-effective and environmentally friendly. These findings provide valuable insights for developing sustainable LIB recycling technologies
MobileViT-based Detection of Anomaly in Measurements of Nuclear Power Plant Core Temperature
No full textInternational audienceThis paper presents a simple model based on MobileViT-v2 for temperature monitoring within a nuclear power plant. Specifically, it is proposed to use MobileNet-v2 to detect a critical accident: a total and instantaneous blockage. On the one hand, the temperature effects of such an event is modeled ans a MobileViT-v2 model for detection. The trained classifier's results are then used in a sequential procedure to detect blockage as quickly and reliably as possible. We compare the performance of two sequential detection methods, namely slidingwindow and CUSUM, in terms of mean detection delay and probability of detection before a prescribed maximum detection delay. Experimental results, using actual temperature measurements from the Superphénix power station, demonstrate the effectiveness of the proposed detection method.</div
Avancement du recyclage des polymères renforcés de fibres de verre issus des pales d’éoliennes en fin de vie grâce à un nouveau processus de délamination.
No full textInternational audienc
Sparse inference in Poisson Log-Normal model by approximating the L0-norm
No full textInternational audienceVariable selection methods are required in practical statistical modeling, to identify and include only the most relevant predictors, and then improving model interpretability. Such variable selection methods are typically employed in regression models, for instance in this article for the Poisson Log Normal model (PLN, Chiquet et al., 2021). This model aim to explain multivariate count data using dependent variables, and its utility was demonstrating in scientific fields such as ecology and agronomy. In the case of the PLN model, most recent papers focus on sparse networks inference through combination of the likelihood with a L1 -penalty on the precision matrix. In this paper, we propose to rely on a recent penalization method (SIC, O'Neill and Burke, 2023), which consists in smoothly approximating the L0-penalty, and that avoids the calibration of a tuning parameter with a cross-validation procedure. Moreover, this work focuses on the coefficient matrix of the PLN model and establishes an inference procedure ensuring effective variable selection performance, so that the resulting fitted model explaining multivariate count data using only relevant explanatory variables. Our proposal involves implementing a procedure that integrates the SIC penalization algorithm (epsilon-telescoping) and the PLN model fitting algorithm (a variational EM algorithm). To support our proposal, we provide theoretical results and insights about the penalization method, and we perform simulation studies to assess the method, which is also applied on real datasets
Le codage de l’information médicale à l’épreuve de l’IA.: Performance et incertitude du codage en centre hospitalier
No full textInternational audienceThe deployment of artificial intelligence (AI) in healthcare organizations is not limited to the fields of research and care; the encoding of medical data, an essential part of hospital administration, is also challenged by AI, in the framework of activity-based pricing. Drawing on an interview and observation survey in a French hospital, we analyse the methods and effects of introducing AI into a Medical Information Department. While AI currently has only a marginal impact on professional and communication practices and procedures within this department, we show that the main reason for integrating AI is the financial efficiency of encoding in the context of activity-based pricing. The article establishes not only that AI does not reduce uncertainty in the encoding process, but also that it is likely to generate uncertainty about the quality of the encoded data, whereas it is an essential resource for medical research.Le déploiement de l’intelligence artificielle (IA) dans les organisations de santé ne se limite pas aux domaines de la recherche et du soin. Le codage de l’information médicale, étape administrative essentielle au fonctionnement hospitalier, en particulier dans le cadre de la tarification à l’activité, en est également le théâtre. À partir d’une enquête menée par entretiens et observations au sein d’un centre hospitalier français, nous analysons les modalités et effets de l’introduction de l’IA au sein d’un Département d’information médicale (DIM). Alors que l’IA n’a actuellement qu’un impact marginal sur les pratiques et modalités de communications professionnelles au sein de ce DIM, nous montrons que le motif principal d’intégration du dispositif est l’efficience financière du codage dans le cadre de la tarification à l’activité. L’article établit non seulement que l’IA ne réduit pas l’incertitude relative au codage mais aussi qu’elle est susceptible de générer des incertitudes inhérentes à la qualité des données produites, alors même que ces dernières constituent des ressources indispensables à la recherche médicale
Integration of bright color centers into arrays of silicon carbide nanopillars
No full textSilicon carbide (SiC) is a wide-bandgap semiconductor combining mature fabrication processes with the ability to host optically active point defects, called color centers, making it ideal for quantum technologies. This study focuses on silicon vacancy defects in 4H-SiC, integrated into nanopillar arrays fabricated via ion implantation, e-beam lithography, and reactive ion etching. SiC nanopillars with a height of 1.4µm are formed and arrays with varying pillar diameters and spacings are obtained. Cathodoluminescence measurements at 80 K reveal a 2–4 times improvement in light collection efficiency from the defects compared to unstructured SiC. The cathodoluminescence intensity increases with a smaller pillar spacing and a larger diameter. These findings demonstrate the potential of SiC nanopillar arrays as scalable platforms for enhancing quantum photonic device performance
Efficient Distance Pruning for Process Suffix Comparison in Prescriptive Process Monitoring
No full textInternational audiencePrescriptive process monitoring seeks to recommend actions that improve process outcomes by analyzing possible continuations of ongoing cases. A key obstacle is the heavy computational cost of large-scale suffix comparisons, which grows rapidly with log size. We propose an efficient retrieval method exploiting the triangle inequality: distances to a set of optimized pivots define bounds that prune redundant comparisons. This substantially reduces runtime and is fully parallelizable. Crucially, pruning is exact: the retrieved suffixes are identical to those from exhaustive comparison, thereby preserving accuracy. These results show that metric-based pruning can accelerate suffix comparison and support scalable prescriptive systems
