Search

International audience; Challenging times for European electricity security have recently brought light to the importance of a robust power sector with well-functioning electricity markets. Increased uncertainty disrupts the standard practice of decision-making in energy systems. The increased penetration of Renewable Energy Sources (RES) such as wind and photovoltaic plants adds to this uncertainty due to the weather dependency of their electricity production. Artificial Intelligence (AI) based tools have proven their efficiency in different applications in the energy sector ranging from forecasting to optimization and decision making. They permit to simplify modeling chains and to improve performance due to higher learning capabilities compared to state-of-the-art methods. However, decision-makers of the energy sector, especially in high-risk situations, need to understand how decision-aid tools construct their outputs from the data. AI-based tools are often seen as black-box models and this penalizes their acceptability by end-users (traders, power system operators a.o.). The lack of interpretability of AI tools is a major challenge for the wider adoption of AI in the energy sector and a fundamental requirement to better support humans in the decision-aid process. Agents of energy systems expect very high levels of reliability for the various services they provide. As energy systems are impacted by multiple uncertainty sources (e.g. available power of RES plants, climate change, market conditions), developed AI tools should not only be performant on average situations but be able to guarantee robust solutions in the case of extreme events. Therefore, our research focuses on optimizing understandable symbolic representations of data-driven decision-aid models for human operators in the energy sector.

We introduce an Alternating Direction Method of Multipliers (ADMM) for finding a solution of the nonsymmetric Eigenvalue Complementarity Problem (EiCP). A simpler version of this method is proposed...

The use of automotive air conditioning leads to a fuel overconsumption. To reduce this overconsumption, we can either work upstream on the technical definitions of the cabin and the HVAC system or optimize control strategies. In both cases, it is essential to build a cabin thermal model that well balances accuracy and complexity. This is the topic of this PhD thesis driven by Renault Group. First, a model reduction methodology is used to build a 0D model starting from a 3D finite element cabin thermal model. This 0D model is based on mass and energy balances on the different cabin walls and air zones. It consists of a nonlinear differential algebraic equations system which can be reinterpreted as a Bond Graph. In addition, the 0D model is based on a weak coupling between the thermal equations and the fluid mechanics ones resulting from CFD calculations (internal airflow and external aerodynamics). Secondly, we apply a machine learning method to the data generated by the 0D model in order to build a reduced 0D model. A design of experiment is considered at this stage. Due to the nonlinearity of the heat exchanges, we have developed an approach which is inspired by the Gappy POD and EIM methods. We use a multiphysics reduced basis that takes several contributions into account (temperatures, enthalpies, heat fluxes and humidities). The resulting reduced model is a hybrid model that couples some of the original physical equations to an artificial neural network. The reduction methodology has been validated on Renault vehicles. The reduced order models have been integrated into a vehicle system-level energetic simulation platform (GREEN) which models different thermics (engine, transmission, cooling system, battery, HVAC, refrigerant circuit, underhood) in order to perform thermal management studies which are of particular importance for electric and hybrid vehicles. The reduced order models have been validated on several scenarios (temperature control for thermal comfort, driving cycles, HVAC coupling) and have achieved CPU gains of up to 99% with average errors of 0.5 °C on temperatures and 0.6% on relative humidities.; L’utilisation de la climatisation automobile engendre physiquement une surconsommation de carburant. Pour diminuer cette surconsommation, il existe deux leviers principaux. Le premier consiste à travailler en amont sur la définition technique de l’habitacle et du système de climatisation. Le second levier consiste à optimiser les stratégies de contrôle. Dans les deux cas, il s’avère incontournable de construire des modèles de thermique habitacle précis et rapides à évaluer. Ce qui fait l’objet de cette thèse CIFRE du Groupe Renault. Dans un premier temps, une méthodologie de réduction de modèles est exploitée pour passer d’un modèle éléments finis 3D à un modèle 0D. Ce modèle 0D est basé sur des bilans de masse et d’énergie sur les différentes parois et zones d’air de la cabine. Il prend la forme d’un système d’équations algébro-différentielles non-linéaire qui peut être transcrit en Bond Graph. De plus, le modèle 0D exploite un couplage faible entre la thermique et la mécanique des fluides issue des calculs CFD (aéraulique et aérodynamique externe). Dans un deuxième temps, on applique une méthode d’apprentissage automatique aux données générées par le modèle 0D en vue de construire un modèle 0D réduit. Un plan d’expériences est considéré à cette étape. Du fait de la non-linéarité des échanges thermiques, nous avons développé une approche qui s’inspire des méthodes Gappy POD et EIM. La base réduite utilisée est une base multiphysique qui tient compte de plusieurs contributions (températures, enthalpies, flux thermiques et humidités). Le modèle réduit obtenu est un modèle hybride qui couple quelques équations physiques d’origine à un réseau de neurones artificiel. La méthodologie de réduction a été déployée sur des véhicules Renault. Les modèles réduits ont été intégrés dans la plateforme GREEN de synthèse énergétique qui modélise différentes thermiques (moteur, transmission, circuit de refroidissement, batterie, HVAC, boucle froide, sous-capot) en vue de faire des études de gestion thermique qui revêtent une importance particulière pour les véhicules électriques et hybrides. Les modèles réduits ont été validés sur plusieurs scénarios (boucle de régulation pour le confort thermique, cycle d’homologation, couplage HVAC) et ont permis d’obtenir des gains CPU allant jusqu’à 99% avec des erreurs moyennes de 0,5°C sur les températures et 0,6% sur les humidités relatives.

The use of automotive air conditioning leads to a fuel overconsumption. To reduce this overconsumption, we can either work upstream on the technical definitions of the cabin and the HVAC system or optimize control strategies. In both cases, it is essential to build a cabin thermal model that well balances accuracy and complexity. This is the topic of this PhD thesis driven by Renault Group. First, a model reduction methodology is used to build a 0D model starting from a 3D finite element cabin thermal model. This 0D model is based on mass and energy balances on the different cabin walls and air zones. It consists of a nonlinear differential algebraic equations system which can be reinterpreted as a Bond Graph. In addition, the 0D model is based on a weak coupling between the thermal equations and the fluid mechanics ones resulting from CFD calculations (internal airflow and external aerodynamics). Secondly, we apply a machine learning method to the data generated by the 0D model in order to build a reduced 0D model. A design of experiment is considered at this stage. Due to the nonlinearity of the heat exchanges, we have developed an approach which is inspired by the Gappy POD and EIM methods. We use a multiphysics reduced basis that takes several contributions into account (temperatures, enthalpies, heat fluxes and humidities). The resulting reduced model is a hybrid model that couples some of the original physical equations to an artificial neural network. The reduction methodology has been validated on Renault vehicles. The reduced order models have been integrated into a vehicle system-level energetic simulation platform (GREEN) which models different thermics (engine, transmission, cooling system, battery, HVAC, refrigerant circuit, underhood) in order to perform thermal management studies which are of particular importance for electric and hybrid vehicles. The reduced order models have been validated on several scenarios (temperature control for thermal comfort, driving cycles, HVAC coupling) and have achieved CPU gains of up to 99% with average errors of 0.5 °C on temperatures and 0.6% on relative humidities.; L’utilisation de la climatisation automobile engendre physiquement une surconsommation de carburant. Pour diminuer cette surconsommation, il existe deux leviers principaux. Le premier consiste à travailler en amont sur la définition technique de l’habitacle et du système de climatisation. Le second levier consiste à optimiser les stratégies de contrôle. Dans les deux cas, il s’avère incontournable de construire des modèles de thermique habitacle précis et rapides à évaluer. Ce qui fait l’objet de cette thèse CIFRE du Groupe Renault. Dans un premier temps, une méthodologie de réduction de modèles est exploitée pour passer d’un modèle éléments finis 3D à un modèle 0D. Ce modèle 0D est basé sur des bilans de masse et d’énergie sur les différentes parois et zones d’air de la cabine. Il prend la forme d’un système d’équations algébro-différentielles non-linéaire qui peut être transcrit en Bond Graph. De plus, le modèle 0D exploite un couplage faible entre la thermique et la mécanique des fluides issue des calculs CFD (aéraulique et aérodynamique externe). Dans un deuxième temps, on applique une méthode d’apprentissage automatique aux données générées par le modèle 0D en vue de construire un modèle 0D réduit. Un plan d’expériences est considéré à cette étape. Du fait de la non-linéarité des échanges thermiques, nous avons développé une approche qui s’inspire des méthodes Gappy POD et EIM. La base réduite utilisée est une base multiphysique qui tient compte de plusieurs contributions (températures, enthalpies, flux thermiques et humidités). Le modèle réduit obtenu est un modèle hybride qui couple quelques équations physiques d’origine à un réseau de neurones artificiel. La méthodologie de réduction a été déployée sur des véhicules Renault. Les modèles réduits ont été intégrés dans la plateforme GREEN de synthèse énergétique qui modélise différentes thermiques (moteur, transmission, circuit de refroidissement, batterie, HVAC, boucle froide, sous-capot) en vue de faire des études de gestion thermique qui revêtent une importance particulière pour les véhicules électriques et hybrides. Les modèles réduits ont été validés sur plusieurs scénarios (boucle de régulation pour le confort thermique, cycle d’homologation, couplage HVAC) et ont permis d’obtenir des gains CPU allant jusqu’à 99% avec des erreurs moyennes de 0,5°C sur les températures et 0,6% sur les humidités relatives.

International audience; A monolithic Lagrangian finite element method is explored to model crack propagation. This approach is based on the level-set method coupled with anisotropic remeshing to define the crack faces and tip. Moreover, the Gθ method is used for computing the strain energy release rate and the propagation direction. Furthermore, a new technique for computing the convection velocity, which is a mixture between the propagation speed and the mechanical velocity, is introduced.

Climate change will impact renewable resources and electricity demand, usually not jointly considered when designing future decarbonized power systems. This paper assesses how sensitive the Portuguese carbon-neutral power sector is to climate change by 2050 and what are the implications for the formally approved Portuguese Carbon Neutrality Roadmap. The future capacity factors for wind, solar and hydropower and electricity demand response to temperature are estimated for 22 climate projections along the Representative Concentration Pathway 4.5 and 8.5. The eTIMES_PT optimization model is used to assess its combined impact on the cost-optimal configuration of the power sector by 2050. Results show that climate change lowers hydropower generation by 20% (in median terms). Improving spatial and temporal resolution and including future climate patterns, results also in lower cost-effectiveness of solar photovoltaic vis-a-vis the Carbon Neutrality Roadmap. While future climate does not impact onshore wind production, offshore wind power generation is positively affected, being a climate-resilient carbon-neutral option for Portugal. Annual electricity unitary costs at final users (excluding taxes and levies) only increase up to 4% with climate change, but seasonal costs have higher variability. This analysis highlights that climate change affects the cost-optimal annual carbon-neutral power sector and needs to be included in energy planning. info:eu-repo/semantics/publishedVersion

Le patrimoine culturel matériel est le fruit des savoir-faire de haute technicité, constitués et transmis au fil des siècles au sein des communautés locales grâce à l’intelligence du geste associée à la créativité de l’esprit humain. Or ce sont ces mêmes savoir-faire qui aujourd’hui sont menacés par la mondialisation et l’industrialisation, du fait notamment de la dévalorisation des métiers manuels et de l’impossibilité jusqu’à ce jour de les « mettre en boîte », de les enregistrer et de les codifier de manière à pouvoir les transmettre aux nouvelles générations. Cet article présente un aperçu des outils et des méthodes actuels, qui vont de l’ethnographie numérique au multimédia, et propose une nouvelle stratégie de transmission de ce savoir-faire basée sur les technologies de reconnaissance du geste technique. The tangible cultural heritage is the result of high level know-how, acquired and transmitted over centuries within local communities through the association of gestural intelligence and the creativity of the human spirit. However, this know-how is currently threatened by globalization and industrialization, due in particular to the devaluation of manual work and the difficulty of “boxing up” this gestural expertise, of recording and codifying it so that it can be transmitted across generations. This article presents an overview of existing methods and tools and proposes a new strategy for know-how transmission based on motion capture and gesture recognition technologies. El patrimonio cultural material es el resultado de una sabiduría de alto nivel, adquirida a lo largo de los siglos y que se transmite dentro de las comunidades locales gracias a la inteligencia del gesto asociada a la creatividad del espíritu humano. Pero esta sabiduría es amenazada hoy a causa de la globalización y la industrialización, debido en particular a la devaluación de la artesanía manual y la imposibilidad hasta ahora de salvaguardar esta experiencia gestual, grabándola y la codificándola de modo que pueda ser transmitida a las nuevas generaciones. En esta sección se presenta una visión general de los métodos y herramientas existentes y se propone una nueva estrategia para la transferencia de esa sabiduría basada en “captura de movimiento” (motion capture) y tecnología de reconocimiento de gestos.

We propose in this work an adaptive Eulerian framework for the simulation of both boiling and evaporation phenomena occurring at the interface of a heated 3D solid immersed in a liquid tank. It simultaneously takes into account the gas-liquid phase changes, the vapor formation and their dynamics, and consequently the 3D quenching or cooling of a heated solid. It uses a Level Set method to separate and to track each phase. The phase change is performed using the balance of heat fluxes at the interface without the use of conforming mesh. Instead, the use of an a posteriori error estimate leading to highly stretched anisotropic elements at the interface enables to drastically reduce errors on computed jumps. This avoids the need of interface reconstruction or interpolation procedure. Finally, a Variational Multiscale solver for the Navier-Stokes equations is extended with implicit treatment of the surface tension. A series of 2D and 3D problems are solved to highlight the efficiency and the accuracy of the proposed framework. The cooling of an immersed solid is also presented and shows good agreement with experimental data. To the best of our knowledge, direct numerical simulations of quenching using an Eulerian framework with boiling and evaporation have never been considered.

International audience; A spectral algorithm is proposed to simulate an isotropic Gaussian random field on a sphere equipped with a geodesic metric. This algorithm supposes that the angular power spectrum of the covariance function is explicitly known. Direct analytic calculations are performed for exponential and linear covariance functions. In addition, three families of covariance functions are presented where the calculation of the angular power spectrum is simplified (shot-noise random fields, Yadrenko covariance functions and solutions of certain stochastic partial differential equations). Numerous illustrative examples are given.

International audience; The full numerical simulation of quenching process requires tackling several phenomena: multiphase flows, fluid structure interaction and wetting, phase change, heat exchanges, etc. On the one hand, an adaptive Eulerian framework for the simulation of boiling phenomena that captures the most part is presented. However no solid-liquid contacts in parallel with boiling is allowed. On the other hand, a discontinuous Galerkin finite element method applied to the coupled unsteady Stokes/Cahn-Hilliard equations is presented. This framework provides simulations of wetting phenomena on simple cases. The opportunity to combine both frameworks for a full simulation of quenching process is discussed.

International audience; The book covers the main mining issues where geostatistics, a discipline founded in the 1960s to study regionalised variables measured at a limited number of points in space, is expected to play a role, from geological interpretation and modelling to exploitation and grade control, evaluation of in situ resources and recoverable resources, long-term mine planning, geotechnics, geometallurgy and sampling. The presentation, with more than 150 illustrations, is systematically oriented towards a practical exposure of the problems and proposed solutions and is illustrated through case studies in Chilean copper porphyry deposits. It is the result of the work published by the authors, researchers from Mines ParisTech and the University of Chile, made in collaboration with the Chilean company Codelco since the late 1990s and early 2000s. The range of problems addressed and their practical presentation make this book unique and usable by a very wide audience - students, researchers, geologists, engineers, geotechnicians and metallurgists.; L’ouvrage couvre les principales problématiques minières où la géostatistique, discipline fondée dans les années 1960 pour étudier les variables mesurées en un nombre limité de points de l’espace, joue un rôle essentiel. Sont concernés l’interprétation et la modélisation géologique, l’exploitation et le contrôle des teneurs, l’évaluation des ressources, la planification minière, la géotechnique, la géométallurgie et l’échantillonnage. Riche de plus de 150 illustrations, la présentation est systématiquement orientée vers une exposition pratique des problèmes posés par l’industrie et des solutions mises en œuvre. Transposables à d’autres types de gisements, les cas concrets concernent ici des gisements porphyriques cuprifères chiliens, études publiées par les auteurs et réalisées en collaboration avec l’entreprise chilienne Codelco depuis la fin des années 1990. L’éventail des problèmes abordés et leur présentation pratique rendent l’ouvrage utilisable par un large public – entre autres étudiants, chercheurs, géologues, ingénieurs, géotechniciens et métallurgistes.

Most surface 3D scanning techniques produce 3D point clouds. This thesis tackles the problem of using points as only explicit surface representation. It presents two contributions in point-based processing.The first contribution is a new raw and massive point cloud screen-space rendering algorithm. This new method can be applied to a wide variety of data from small objects to complex scenes. A sequence of screen-space pyramidal operators is used to reconstruct in real-time a surface and estimate its normals, which are later used to perform deferred shading. In addition, the use of pyramidal operators allows to achieve framerate one order of magnitude higher than state of the art methods.The second proposed contribution is a new immersed boundary computational fluid dynamics method by extended implicit surface reconstruction. The proposed method is based on a new implicit surface definition from a point cloud by extended moving least squares. This surface is then used to define the boundary conditions of a finite-elements immersed boundary transient Navier-Stokes solver, which is used to compute flows around the object sampled by the point cloud. The solver is interfaced with an anisotropic and adaptive meshing algorithm which refines the computational grid around both the geometry defined by point cloud and the flow at each timestep of the simulation.; Le nuage de points 3D est la donnée obtenue par la majorité des méthodes de numérisation surfacique actuelles. Nous nous intéressons ainsi dans cette thèse à l’utilisation de nuages de points comme unique représentation explicite de surface. Cette thèse présente deux contributions en traitements basés points.La première contribution proposée est une nouvelle méthode de rendu de nuages de points bruts et massifs par opérateurs pyramidaux en espace image. Cette nouvelle méthode s’applique aussi bien à des nuages de points d’objets scannés, que de scènes complexes. La succession d’opérateurs en espace image permet alors de reconstruire en temps réel une surface et d’en estimer des normales, ce qui permet par la suite d’en obtenir un rendu par ombrage. De plus, l’utilisation d’opérateurs pyramidaux en espace image permet d’atteindre des fréquences d’affichage plus élevées d’un ordre de grandeur que l’état de l’art.La deuxième contribution présentée est une nouvelle méthode de simulation numérique en mécanique des fluides en volumes immergés par reconstruction implicite étendue. La méthode proposée se base sur une nouvelle définition de surface implicite par moindres carrés glissants étendue à partir d’un nuage de points. Cette surface est alors utilisée pour définir les conditions aux limites d’un solveur Navier-Stokes par éléments finis en volumes immergés, qui est utilisé pour simuler un écoulement fluide autour de l’objet représenté par le nuage de points. Le solveur est interfacé à un mailleur adaptatif anisotrope qui permet de capturer simultanément la géométrie du nuage de points et l’écoulement à chaque pas de temps de la simulation.

International audience

International audience

International audience

International audience

International audience

International audience

Intangible Cultural Heritage (ICH) creations include, amongst other, music, dance, singing, theatre, human skills, and craftsmanship. These cultural expressions are usually transmitted orally and/or using gestures and are modified over a period of time, through a process of collective recreation. As the world becomes more interconnected and many different cultures come into contact, local communities run the risk of losing important elements of their ICH, while young people find it difficult to maintain the connection with the cultural heritage treasured by their elders. In this paper, we present a novel holistic approach for the safeguarding and transmission of ICH that goes beyond the mere digitization of ICH content. Based on multisensory technology for the capturing of ICH, the proposed approach enables the generation of completely novel cultural content. High-level semantics are extracted from the acquired data, enabling researchers to identify possible implicit or hidden correlations between different ICH expressions or interpretation styles and study the evolution of a specific ICH. These data, coupled with other cultural resources, are accessible through the i-Treasures Web-platform, which provides the means for supporting knowledge exchange between researchers as well as know-how transmission from ICH bearers to apprentices.

International audience; This paper addresses the drone tracking problem, using a model based on the Frenet-Serret frame. A kinematic model in 2D, representing intrinsic coordinates of the drone is used. The tracking problem is tackled using two recent filtering methods. On the one hand, the Invariant Extended Kalman Filter (IEKF), introduced in [1] is tested, and on the other hand, the second step of the filtering algorithm, i.e. the update step of the IEKF is replaced by the update step of the Unscented Kalman Filter (UKF), introduced in [2]. These two filters are compared to the well known Extended Kalman Filter. The estimation precision of all three algorithms are computed on a real drone tracking problem.

In this paper we consider a noise free class of dynamics encompassing left- and right-invariant on a Lie group with noisy partial state measurements. We assume in addition that the covariance matrix of the state is initially rank-deficient. This, combined with the absence of process noise, keeps the system state within a (time-dependent) subset of the state space at all times. We prove mathematically that the invariant extended Kalman filter (IEKF) perfectly respects this kind of state constraints, contrarily to the standard EKF, or the unscented Kalman filter. This is a strong indication that the IEKF is particularly well suited to navigation when motion sensors are highly precise. The theory is applied to a non-holonomic car example on SE (2), and to an attitude estimation example on SO(3).

In this paper we consider the problem of tracking a maneuvering aircraft. The dynamical model used is based on assumptions of nearly constant tangential velocity, curvature, and torsion, of the target trajectory. Using the Frenet-Serret frame, that is, an element of the special orthogonal group SO(3) we cast the tracking problem into a filtering on Lie groups framework. We then use an invariant extended Kalman filter to estimate the various quantities involved. The resulting filter is simple to implement, and shown to gracefully accomodate some realistic target trajectories.

International audience; In this paper we consider the problem of tracking a target in a 2D space whose model is based on a constant velocity assumption (up to a process noise) in the Frenet-Serret frame (or intrinsic coordinates). This model is particularly suited to straight lines and coordinated turn motions. Then, we adapt the Invariant Extended Kalman Filter (I-EKF), as in [1], a novel variant of the EKF on Lie groups, to suit this dynamical model. Numerical experiments inspired by real fighter trajectories confirm the validity of our approach.

Sample measurements (of grade, depth, etc.) are almost inevitably affected by errors. Several error models were studied in the literature. But the interest of replicates for selecting the error model received limited attention. If measurement errors are supposed to be additive, homoscedastic, without correlation between them, and spatially not correlated with the exact values, the variances of the measurement errors are computable from the sample, simple, and cross-variograms of replicate data sets, even if the variogram of the exact value is pepitic (Aldworth W, Spatial prediction, spatial sampling, and measurement error. Retrospective Theses and Dissertations. Paper 11842. Iowa State University Digital Repository @ Iowa State University, 1998; Faucheux et al. Characterisation of a hydrocarbon polluted soil by an intensive multi-scale sampling. Geostats 2008, proceedings of the 8th international geostatistics congress, 1–5 Dec. 2008, Santiago, Chile. Ortiz J-M, Emery X (eds) for an example, 2008). But what about the other cases? When the error is additive, its correlation with the exact value can remain undetectable. The variance of the measurement errors is thus not always computable. It’s the same for an error of multiplicative type. Except in some special cases, keeping the different measurement values rather than their average improves the precision of the estimation.

International audience; An analytical model for growth in a semi-infinite matrix with cross-diffusion between species is presented. Application is given for precipitation of the-phase in the-matrix during isothermal holding at 600 °C in the Ni – 7.56 at.% Al – 8.56 at.% Cr alloy. The exact time-dependent solutions for the solute profiles and the growth kinetics are validated with a numerical front-tracking simulation. The simulation of cross diffusion terms in a multicomponent alloy is thus demonstrated. Extension of the analytical solution is given for growth in a matrix of finite size. The driving force is then based on a mathematical estimation of the far-field composition. The Gibbs-Thomson effect is also accounted for to consider the effect of curvature on the equilibrium tie-lines. Comparison of analytical solution with the numerical front-tracking simulation shows excellent agreement. Results also point out the detrimental approximation of using the average composition of the matrix for computing the driving force as well as the limitation of the solution proposed by Chen et al. [Acta Mater. 56 (2008) 1890]. A detailed discussion is finally given on the origin of oscillations observed for the time evolution of the precipitate radius which alternates between growth and dissolution regimes, pointing out the combined role of solute fluxes and tie-lines compositions at the precipitate/matrix interface.

International audience; The following is a geostatistical study of copper measurements on samples from diamond drill-holes and blast-holes. Both measurements are formally compared, leading to a model where a blast-hole can be considered a regularization of the drill information up to a nugget effect characteristic of the blast-holes. This formal link makes it possible to build a cokriging system that takes into account the different supports and leads to a block model based on blast and drill-holes. The model is tested on a realistic simulation where the true block grades, which are known, are compared to their estimate obtained by:-Kriging using only drill-holes;-Kriging using only blast-holes;-Cokriging using drill and blast-holes together. A preliminary conclusion is that the best estimates are obtained when only blasts or alternatively, blast and drill-holes are used; there is no significant difference between the two, which is due to the great amount of blast information. This result justifies the usual practice of basing short-term planning on blasts only. But another conclusion may be drawn when kriging is compared to a moving average (another common practice), both based on blasts: depending on the number of data used in the neighborhood, the moving average produces a strong conditional bias. As a byproduct, we also show how it is possible to filter the blast error by kriging and to make a deconvolution to estimate point support values using blast measurements .

The new generation of reservoir computer models based on unstructured meshes requires appropriate methods for filling the cells with petrophysical data. Classical methods are not applicable to unstructured reservoir models since they can neither treat complicated geometry nor reproduce correctly the statistical properties of the model (as marginal distributions and covariance between the blocks). In order to overcome these difficulties, we develop new statistical methods for model filling which work directly on unstructured grids. The most important problem in the model filling is the transfer of the small scale data available from well cores and laboratory analysis to multiple different scales which are defined by various grid blocks. This problem is known in as the volume support effect and can be addressed by the methods of non-linear geostatistics. In the proposed solution, we use non-linear geostatistics to address the problem of simulating heterogeneities on unstructured grids (Zaytsev et al., 2015). We present a workflow for approximating the block distributions by the means of the discrete Gaussian model and for performing geostatistical simulations directly on block support. Theoretical model and case studies are presented.

Sparse representation of seismic data is a crucial step for seismic forward modeling and seismic processing such as coherent noise separation, imaging, and sparsity-promoting data recovery. In this letter, a new locally adaptive wavelet transform, called easy-path wavelet transform (EPWT), is applied for the sparse representation of seismic data. The EPWT is an adaptive geometric wavelet transform that works along a series of special pathways through the input data and exploits the local correlations of the data. The transform consists of two steps: reorganizing the data following the pathways according to the data values and then applying a 1-D wavelet transform along the pathways. This leads to a very sparse wavelet representation. In comparison to conventional wavelets, the EPWT concentrates most of the energy of signals at smooth scales and needs less significant wavelet coefficients to represent signals. Numerical experiments show that the new method is really superior over the conventional wavelets and curvelets in terms of sparse representation and compression of seismic data.

International audience; The methods of constrained approximation in Hilbert spaces of analytic functions are applied to the solution of the inverse problems of detecting cracks or sources in a two-dimensional material by means of boundary measurements. Issues of well-posedness are discussed, and results on continuity and robustness with respect to the given data are established. Constructive and efficient methods for resolution of the above approximation problems are presented. The techniques are illustrated by numerical examples incorporating a further rational approximation step.

International audience; Predictions of wind power production for horizons up to 48-72 h ahead comprise a highly valuable input to the methods for the daily management or trading of wind generation. Today, users of wind power predictions are not only provided with point predictions, which are estimates of the conditional expectation of the wind generation for each look-ahead time, but also with uncertainty estimates given by probabilistic forecasts. In order to avoid assumptions on the shape of predictive distributions, these probabilistic predictions are produced from non-parametric methods, and then take the form of a single or a set of quantile forecasts. The required and desirable properties of such probabilistic forecasts are defined and a framework for their evaluation is proposed. This framework is applied for evaluating the quality of two statistical methods producing full predictive distributions from point predictions of wind power. These distributions are defined by a number of quantile forecasts with nominal proportions spanning the unit interval. The relevance and interest of the introduced evaluation framework are discussed.

International audience; The El Teniente mine is famous not only as one of the largest known porphyry-copper ore bodies but also, among geologists, for its typical breccia pipe named “Braden”, an almost vertical poorly mineralized cone, located at the center of the mine and surrounded by early-stage mineralizations. As the edge of the pipe constitutes the limit of the deposit and of the mining operation, estimating it accurately is important. In this paper, we are interested in estimation of the elevation of the pipe surface using a geostatistical approach based on non-stationary covariances. Previous approaches have been applied on this dataset by Séguret and Celhay (2013). The proposed estimation method offers an integrated treatment of all aspects of non-stationarity (mean, variance, spatial continuity) in the modelling process. The proposed method has revealed an increased prediction accuracy when compared to standard stationary method, and demonstrated the ability to extract the underlying non-stationarity from a single realization. A comparison of predictions and prediction standard deviations maps indicates that the proposed non-stationary method captures some varying spatial features (such as locally varying anisotropy) in the data that are not present using the stationary method, the outcome appears more realistic.

Les machines multiprocesseurs, multi-cœurs et les accélérateurs de type GPU se généralisent et pourtant il devient de plus en plus difficile pour les programmeurs de tirer profit de leurs capacités. La compilation source-à-source des applications permet de faciliter le développement d’implémentations efficaces pour ces architectures complexes.Mes travaux de recherche s’inscrivent selon deux axes principaux. Le premier relève de la compilation et de l’optimisation d’applications en vue de leur exécution efficace sur des architectures parallèles. Le deuxième axe relève de l’utilisation de l’algèbre linéaire en nombres entiers pour modéliser les problèmes rencontrés.Cette thèse présente, tout d’abord, les analyses statiques et dynamiques de programmes développées pour vérifier la correction d’un code et faciliter sa maintenance, telles que la mise en conformité par rapport à la norme du langage source, la détection de variables non initialisées, la vérification du non-débordement des accès à des éléments de tableaux et le calcul de dépendances de données. Ces analyses ont été intégrées dans le compilateur PIPS.Puis, des méthodes de génération automatique de code de contrôle et des communications à partir de spécifications sont exposées. Des modélisations du placement des données sur les processeurs, des calculs pouvant être exécutés en parallèle et des communications devant être générées pour conserver la cohérence des données sont proposées. Les algorithmes de synthèse de code spécifiquement développés pour HPF, pour une mémoire virtuelle partagée émulée sur une architecture distribuée et pour des transferts compatibles avec des DMAs sont ensuite détaillées.Finalement, une approche globale du problème de placement d’une application pour une architecture embarquée parallèle, qui utilise la programmation logique concurrente par contraintes comme moteur, est présentée. Une même abstraction a été choisie pour les analyses et la modélisation des problèmes d’optimisation et de génération de code : un système de contraintes linéaires où les variables sont des entiers. Le choix des Z -polyèdres a permis de rester dans un cadre algébrique disposant d’une large gamme de méthodes de résolution, aisées pour les preuves.

International audience; We pose and solve an extremal problem in the Hardy class H-2 of the disc, involving the best approximation of a function on a subarc of the circle by a H-2 function, subject to a constraint on its imaginary part on the complementary arc. A constructive algorithm is presented for the computation of such a best approximant, and the method is illustrated by a numerical example. The whole problem is motivated by boundary parameter identification problems arising in non-destructive control

International audience; Galilean electromagnetism is derived from a thermodynamic approach. Attention is paid on various regimes allowed by the quasi-static limit. It is emphasized that classical definition of electro-quasi-stationary and magneto-quasi-static regimes does not cover all the low-frequency case studies. As suggested by the thermodynamic approach of electromagnetism, it is proposed to: 1) discuss the validity of the regimes at the device scale and 2) derive the set of equations underlying the regimes from the variations of the magnetic and electrostatic powers instead of the magnitudes of the magnetic flux density and the electric field.

International audience; This article proposes a method improving Kong et al. algorithm called Locally Adaptive Soft-Voting (LASV) algorithm described in " General road detection from a single image ". This algorithm aims to detect and segment road in structured and unstructured environments. Evaluation of our method over different images datasets shows that it is speeded up by up to 32 times and precision is improved by up to 28% compared to the original method. This enables our method to come closer the real time requirements.

110 pages; Course key notes, case study tutorial, papers

The Boolean RF are a generalization of the Boolean RACS. Their construction based on the combination of a sequence of primary RF by the operation supremum or infimum, and their main properties (among which the supremum or infimum infinite divisibility) are given in the case of scalar RF built on a Poisson point process.

157 pages; 1 - Block modeling by partial grades cokriging (stage 4) 2 - Blast and drill holes (stage 1) 3 Element for classification of mineral resource (stage 4) 4 Geotechnics (stage 2) 5 Mathematical morphology and mining

173 pages; Course key notes, case studies, papers

International audience; Although electric field seems useless in non-conducting regions depicted within the quasi-static magnetic regime, its calculation appears necessary therein to: (i) improve the mesh quality required by eddy-current computation; and (ii) tackle the interplay between power transmission and its loss of integrity induced by higher frequencies. The discussion lies on a thermodynamic-oriented description of electromagnetism and a reversible interpretation of the Faraday's law. Energy efficiency challenges deserve such an attention.

International audience; Powder-based microstructures are traditionally modelled using densely packed spherical particles. In this letter, a new dropping and rolling packing algorithm is employed by means of analytical equations, to investigate the random packing of spherical particles. The three-dimensional randomly packed microstructures are quantified in terms of packing density and computational time.

ISBN : 978-3-642-38293-2; International audience; Generalizations of various random tessellation models generated by Poisson point processes are proposed. They are interpreted as characteristics of Boolean random functions models, which provide a generic way of simulation of general random tessellations.

Modelling transport and long-term creep in concrete materials is a difficult problem when the complexity of the microstructure is taken into account, because it is hard to predict instantaneous elastic responses. In this work, several numerical methods are compared to assess their properties and suitability to model concrete-like microstructures with large phase properties contrast. The methods are classical finite elements, a novel extended finite element method (mu-XFEM), an unconstrained heuristic meshing technique (AMIE), and a locally homogenising preprocessor in combination with various solvers (BENHUR). The benchmark itself consists of a number of simple and complex microstructures, which are tested with a range of phase contrasts designed to cover the needs of creep and transport modelling in concrete. The calculations are performed assuming linear elasticity and thermal conduction. The methods are compared in term of precision, ease of implementation and appropriateness to the problem type. We find that XFEM is the most suitable when the mesh if coarse, and methods based on Cartesian grids are best when a very fine mesh can be used. Finite element methods are good compromises with high flexibility. (C) 2013 Elsevier Ltd. All rights reserved.

International audience; Complex microstructures in materials and in biology often involve multi-scale heterogeneous textures, that we model by random sets derived from Mathematical Morphology. Our approach starts from 2D or 3D images; a complete morphological characterization is performed, and used for the identification of a model of random structure. Simulations of realistic microstructures are introduced in a numerical solver to compute appropriate fields (electric, elastic, velocity, ...) and to estimate the effective properties by numerical homogenization, accounting for scale dependent statistical fluctuations of the fields.

ISBN : 978-3-642-21568-1; International audience; Stochastic modeling of a material microstructure is in general composed of multiple steps. First, geometric properties of the sample are measured by image analysis. Second, an appropriate stochastic model is chosen and model parameters are estimated from the geometric properties. Third, additional characteristics are computed on the data set and on realizations of stochastic models to evaluate the quality of the fitting. In this article, we show how to measure geometric properties of a fiber system, estimate parameters for two different fiber models, and evaluate the realizations with orientation covariance and tortuosity of the fibers. The considered stochastic models are the newly developed bended fiber model, composed of a force-biased packing of ball chains, and the classical cherry-pit cylinder model. We show the advantages and limitations of both methods.

International audience; We present a novel method for detecting circles on digital images. This transform is called the circlet transform and can be seen as an extension of classical 1D wavelets to 2D; each basic element is a circle convolved by a 1D oscillating function. In comparison with other circle-detector methods, mainly the Hough transform, the circlet transform takes into account the finite frequency aspect of the data; a circular shape is not restricted to a circle but has a certain width. The transform operates directly on image gradient and does not need further binary segmentation. The implementation is efficient as it consists of a few fast Fourier transforms. The circlet transform is coupled with a soft-thresholding process and applied to a series of real images from different fields: ophthalmology, astronomy and oceanography. The results show the effectiveness of the method to deal with real images with blurry edges.

in press