Portail HAL UPEC
Not a member yet
55480 research outputs found
Sort by
Integrated optimization and machine learning through hyperparameter selection: an application to predictive maintenance of wind turbines
No full textInternational audienceNumerous real-life problems involve two complex challenges: prediction, because of unknown or uncertain parameters or variables, and decision, because of the need to make a good one or the best one. These two challenges are not solved with the same tools: the firstone requires learning and the second one constrained optimization. The literature proposes various ways to link decision and prediction problems. The most common and obvious way is to perform the prediction first and then to optimize the decision based on this prediction. This sequential paradigm is called predict-then-optimize (PtO). Even though this paradigm is widespread, it often leads to sub-optimal decisions becausethe learning model used for the prediction aims at minimizing a loss function on the object to predict and not on the decision that will be taken based on this prediction (2).More recently, authors proposed a new paradigm, called Decision-Focused Learning (DFL), including the decision problem into the prediction problem by changing the loss function of the learning task to minimize the loss on the final decision. It supposes several conditions to be satisfied, because of the need to be able to compute the derivative, or a substitute, of the loss function. This is not immediate when considering a constrained optimization problem, in particular with integer variables (3).The model we propose optimizes the hyperparameter search of the learning model not with a traditional score, such as mean squared error or f1-score, but with a customized score being the objective function of the constrained optimization model of the decision problem. The decision is then integrated into the learning phase. As the scoring function of the hyperparameter search is not subject to the assumptions of derivability, which is the case for the loss function used in DFL, it requires much less effort to integrate the decision and the prediction phases.Predictive maintenance (PdM) is a case involving a decision problem, using constrained optimization, and a prediction problem, using learning. This paper examines how optimization and machine learning can be combined to improve PdM in wind farms. PdM is a maintenance strategy that aims at performing maintenance tasks little before a failure is likely to occur. Being able to schedule a task a little before a failure occurs assumes having information on the future operational state of the equipment. This information can be under the form of remaining useful life, ie. the time remaining before a failure, or of a binary information on the likeliness of a failure occurring the next period of time (4). Having these kinds of forecasts is a prediction problem. Wind turbines are equipped with multiple sensors monitoring and storing various quantities, such as temperatures, angles, rotation speeds, etc. The availability of data makes it suitable to use learning for failures prediction (1).Operation and Maintenance decision-makers wish to elaborate a good maintenance schedule, or even the best one, in the sense that it should minimize the maintenance costs and both the planned and unplanned downtime. The goal is to achieve a compromise between curative maintenance, that systematically causes unplanned downtime and high maintenance costs but guarantees a maximal use of the components, and preventive maintenance, that avoids some of the failures by planning operation at regular time interval with lower maintenance costs but still cause planned downtime, may lead to over-maintaining and does not erase the risk of failures. Predictive tasks should then replace preventive and curativeones.Maintenance planning is thus a decision problem. The maintenance planning is constrained by field rules such as potentiality, security, etc. (5). The tool used to find the best planning while satisfying the filed rules is constrained optimization.In the case of PdM for wind turbines, the constrained optimization program seeks to minimize a cost function composed of maintenance costs and of downtime costs, both planned and unplanned. The decision variables are the beginning of each maintenance task and is a Mixed Integer Linear Program (MILP). The tasks to schedule can be curative, preventive or curative tasks. A predictive task can replace a preventive or curative task if it is scheduled closely enough before this task’s due date. The data we work with to predict the failures is labeled with historical failures, thus thelearning task is not anomaly detection but supervised machine learning. Due to the fact that the failures do not happen as often as normal functioning, the dataset is highly skewed. We use a gradient boosting classification model for the prediction, as it is robust to skewed datasets. This still leads to an excessive amount of false positive, even using the f1-score as scoring function to find the best hyper-parameters, which combines precision and recall to ensure the quality of the prediction of a classifier.This paper proposes a model integrating the failure predictions, using a classification gradient boosting learning algorithm, and the planning optimization, using a scheduling MILP. The integration is done through the hyperparameters selection of the learning model as thescoring function of the grid search is a loss function based on the objective function of the scheduling MILP. Experimentations were carried out on industrial data shared by a partner. We compared the integrated predictive approach to a maintenance strategy purely preventive and curative and to a second maintenance strategy including predictive maintenance according to a PtO model, ie. without the loop on the learning-decision phases. Results show that the integrated model reduces the amount of false negative predictions, as expected. This has the effect of scheduling a number of predictive tasks that is closer to what is in fact needed, replacing preventive tasks and not performing too many predictive tasks. Results are convincing: the integrated predictive approach allows to gain up to 30% compared to the purely preventive and curative strategy and from 10% to 20% compared to the PtO predictive strategy, in maintenance costs
Narrative, visual, and didactic choices in the Supertroupers Project: designing a digital comic on energy
No full textInternational audienceThis paper reports on the design of a digital comic episode for teaching the concept of energy in secondary education. Developed through a collaboration between physics education researchers, a scriptwriter, and an illustrator, the episode introduces energy-related key ideas—such as physical state, transfer, conservation, and sources—through a fictional narrative centred on the construction of an hipotetical energy-generating device. The design was guided by a combined framework of Design-Based Research and the Model of Educational Reconstruction, allowing for alignment between scientific content, student conceptions, and didactic structuring. Classroom testing of a prototype in France and Spain informed several revisions, including clearer sequencing of environmental issues and enhanced narrative coherence. The final resource remains to be tested. It is part of a multilingual series and is accompanied by teacher support materials. This case illustrates the affordances and boundaries of DBR and MER for designing multimodal educational artefacts in international contexts
A semi-exact primal-dual method for steady viscoplastic flows
No full textInternational audienceConstitutive laws for viscoplastic materials involve a multivalued nonlinearity. Duality methods for such equations are known to converge, butthe convergence is slow, requiring a large number of iterations. We introduce here a new iterative method of implicit primal-dual typefor a class of variational problems, with a particular asymmetrical form in terms of the primal and dual unknowns, with an exact resolution of one relation and an approximate resolution of the other (semi-exact method). Two fast algorithms are proposed. The first includes an adaptive choice of a variable parameter, and the second includes a Newton-like linearized correction procedure. Applying the semi-exact method to a steady viscoplastic problem leads at each iteration to the resolution of a Laplace type equation, thus equivalent in terms of cost per iteration to the well-known Augmented Lagrangian or dual FISTA methods. Numerical tests show that the semi-exact method achieves a faster convergencein terms of number of iterations, compared to the augmented Lagrangian method or to the FISTA* method with acceleration, for Bingham or Herschel-Bulkley laws. This is particularly true when a large zero order term is present as is the case when solving a time dependent problem
Motor complications and postural abnormalities interplay in Parkinson's disease
No full textInternational audienceBackground: Postural abnormalities (PA) and motor complications (MCs, including motor fluctuations - MFs- and levodopa-induced dyskinesia - LIDs) are hallmark of Parkinson's disease (PD) progression, yet their relationship remains poorly understood.Objective: To investigate the association between PA and MCs, motor symptoms, and non-motor symptoms (NMS) in patients with PD, and to assess whether PA influences the development of MCs over time.Methods: Data of the prospective NS-Park cohort (27 French PD Expert Centers) were analysed. PA was defined by a score ≥2 on item 3.13 of the MDS-UPDRS-III. Associations between PA and MCs, as well as with other motor symptoms and NMS, were assessed using logistic regression models. We used interval censoring survival models to assess the associations between PA at inclusion and the incidence of MCs. Analyses were adjusted for sex, age, disease duration, dopaminergic dose, and disease severity.Results: Among 13,037 included PD patients (58.7 % male, median age at diagnosis 61 years), 724 (5.6 %) presented with PA. Patients with PA had longer disease duration, higher disease severity, and higher dopaminergic treatment. PA exhibited a higher prevalence of troublesome MFs (OR: 5.96; 95 % CI: 4.25-8.32) and LIDs (OR: 2.81; 95 % CI: 1.79-4.30), while associations with milder MCs were inconsistent. However, PA was not significantly associated with the development of MCs during follow-up.Conclusions: PA are associated with more frequent severe MCs, and a higher burden of motor and NMS, making patient care particularly challenging
Enseignants missionnés dans les institutions culturelles : quelles appropriations des pratiques de médiation pour le public scolaire ?
No full textInternational audienceCommunication donnée le 16 janvier, session modérée par Hyacinthe Rave
De la production à l'exploration : la “danse infernale” de la coopération inter-organisation
No full textInternational audienceL'article « De la production à l'exploration : la “danse infernale” de la coopération inter-organisation » propose une analyse de la série Étoile (Sherman-Palladino & Palladino, 2025) comme terrain d'observation du management dans les organisations artistiques hybrides. À travers l'échange de talents entre le Ballet National de l'Opéra de Paris et le Metropolitan Ballet Theater de New York, il explore comment un dispositif de coopération, d'abord pensé comme un partenariat de production destiné à renforcer la performance financière, glisse progressivement vers un partenariat d'exploration (Segrestin, 2006). Ce glissement transforme la nature des relations inter-organisationnelles et des processus de création, révélant les tensions identitaires, managériales et symboliques propres aux organisations hybrides (Battilana & Dorado, 2010). Ces tensions remettent en question les philosiophies gestionaires initiales de l'échange de talents et contribue à une profonde transformation de la raison d'être des deux compagnies de danse
Involvement of Taiman in juvenile hormone signaling controlling sexual maturation in a male moth
No full textInternational audienceIn insects, juvenile hormone (JH) is essential for orchestrating reproductive events. For example, in the male moth Agrotis ipsilon, the behavioral response to female sex pheromone is linked to neuronal sensitivity in the primary olfactory centers (antennal lobes, ALs), and the maturation of accessory sex glands (ASGs) are known to be age-and JH-dependent. The molecular basis of this regulatory action of JH is not fully deciphered, and we show here that the heterodimerizing partner of Methoprene-tolerant called Taiman (Tai) is essential for the sexual maturation of male A. ipsilon. Tai expression in ALs and ASGs is elevated from the third day of adult life and is responsible for the acquisition of behavioral responsiveness to the sex pheromone and ASG maturation. Tai-deficient old males exhibited altered sexual behavior and delayed ASG maturation. Moreover, the expression levels of Tai and Krüppel homolog 1 (Kr-h1), an early JH-induced transcription factor, were reduced in ALs and ASGs of JH-deprived and Tai-deficient old males, respectively. Exogenous JH injection into young males resulted in precocious sexual maturation and this JH induction was suppressed by Tai silencing. Our results demonstrate that Tai is an actor of the JH signaling pathway that operates in ALs and ASGs to promote pheromone information processing and consequently the display of sexual behavior in synchrony with ASG maturation, ultimately optimizing male reproductive success. Thus, this study provides additional insights into the molecular mechanisms underlying hormonal regulation of sexual maturation in insects
FUJITA-KATO SOLUTIONS AND OPTIMAL TIME DECAY FOR THE VLASOV-NAVIER-STOKES SYSTEM IN THE WHOLE SPACE
No full textWe are concerned with the construction of global-in-time strong solutions for the incompressible Vlasov-Navier-Stokes systemin the whole three-dimensional space. One of our goals is to establish that small initial velocities with critical Sobolev regularity and sufficiently well localized initial kinetic distribution functions give rise to global and unique solutions. This constitutes an extension of the celebrated result for the incompressibleNavier-Stokes equations (NS) that has been established in 1964 by Fujita and Kato. If in addition the initial velocity is integrable, we establish that the total energy of the system decays to 0 with the optimal rate t^{-3/2}, like for the weak solutions of (NS). Our results partly rely on the use of a higher order energy functional that controls the regularity of the velocityand seems to have been first introduced by Li, Shou and Zhang in the contextof nonhomogeneous Vlasov-Navier-Stokes system. In the small data case, we show that this energy functional decays with the rate t^{-5/2}
Monitoring litter typology and loading in stormwater: Implications for waste management and mitigation policies
No full textInternational audienceSewer systems are an overlooked pathway of litter to rivers. We monitored six combined and four separate sewer outlets in France using retention nets over several years to quantify and characterize over 30,000 items > 5 mm, including 91% plastic debris. Total median loading rates were 155 g ha -1 yr -1 (120 g.ha -1 .yr -1 for plastics) and 8 g cap -1 yr -1 (6 g cap -1 .yr -1 for plastics), consistent with the few previous studies available in stormwater and rivers, although litter typology is different. Combined sewers were dominated by sanitary products like wet wipes and towels (46% of plastic mass), while separate systems were dominated by cigarette butts and food packaging related to on-the-go consumption. Sewer monitoring proved complementary to river monitoring, with fewer unidentified items (~15%) and clearer source attribution. Management strongly influenced maintenance costs: internal management and incineration by public authorities was far cheaper (800 € per net and per year) than outsourced landfill disposal (16,000 € per net and per year). Despite operational limits of nets, sewer monitoring provides robust data for litter flux models and for evaluating upstream mitigation policies. Expanding such monitoring, supported by sustained funding and international coordination, will afford reliable and homogeneous method for litter loading estimates into rivers and their source identification:an essential step toward an effective plastics treaty
Rapport 2025 de l'AMF sur le gouvernement d'entreprise des sociétés cotées
No full textInternational audienceAutorité des marchés financiers, Rapport 2025 « Gouvernement d'entreprise et rémunération des dirigeants des sociétés cotées », 12 déc. 202