Suchergebnisse

Open-source revolutionized the software industry through a public, decentralized, and asynchronous development paradigm that fosters collaboration among peers. New practices and stakeholders disrupted the designingprocess, yet led to industrial successes. Due to the digitalization and democratization of the designing process, this approach now spreads to thedevelopment of tangible artifacts. This is open-design.However, open-design currently appears as an umbrella term that encompasses from amateur do-it-yourself projects to sector-scale industrialcollaborations. It is not clear either, how these practices relate to existing designing approaches. Finally, little knowledge about the open-design processis formalized. This impedes the development of adequate tools for helpingpractitioners to make the most of it.Therefore, we investigated how to model the open-design process in thedevelopment of tangible products. First, we developed a typology of open-design practices based on a systematic search and review of the scientificliterature. Then, we selected one of the types identified and modeled thedifferent facets of the designing process (activities carried out, stakeholdersinvolved, and boundary objects used) in this context, using a groundedtheory-based approach.Through our literature review, we mapped open-design in relation to existing designing approaches, and to coined a new definition thereof. Based on624 papers indexed in the Scopus database, we identified three types of practices — do-it-yourself, meta-design, and industrial ecosystem — which arerelated to the status (professional or amateurs) of the processes' stakeholders and addressees. We also constructed two models of the 'do-it-yourselfopen-design' process using semi-directive interviews of 11 project leaderswho took part in the PoC21 innovation camp. They depict open-design as adesigning process influenced by both open-source software developmentand amateur design. We tested the quality of our models and our modelingmethod via statistical ...

Thèse en co direction Science de l'ingénieur et Sciences politiques ; The complexity of urban systems is a constant challenge for the actors who are confronted with it. Prolific multidisciplinary literature shows the appreciation of this phenomenon is widely conditioned by personal culture, disciplinary knowledge and professional know-how. From a research angle, analyzing complex systems from various points of view (user, technician, public authority, researcher) helps generate original development opportunities. This thesis results from a meeting between industrial engineering, political sciences and town planning around a logic of technology transfer. Indeed, the technologies of collaborative design and user-centric innovation are among the answers industrial engineering brings to problems generated by complexity. Our analysis brings us then to consider the potential that the adaptation of distributed collaborative design (DCD) represents. Consequently, we formulate tools, methods and one specific environment necessary for the development of a DCD dedicated to urban projects to facilitate their global and sustainable acceptability. These studies ask subsequently the question of role which should be given to the final user in the urban project design process. Political sciences help us to understand the relations between citizens and experts. This gives us elements to develop a collaborative approach from the conceptualization phase as this impacts the whole development of the projects. More globally, such collaborative processes may contribute to the emergence of sustainable cities ; La complexité des systèmes urbains est un défi constant pour les acteurs qui y sont confrontés. L'appréciation de ce phénomène est largement conditionnée par la posture (usager, technicien, pouvoir public, chercheur) et le bagage culturel, disciplinaire et professionnel, comme en rend compte une littérature pluridisciplinaire prolifique. Sur le plan de la recherche, multiplier les regards sur les systèmes complexes contribue à ...

Thèse en co direction Science de l'ingénieur et Sciences politiques ; The complexity of urban systems is a constant challenge for the actors who are confronted with it. Prolific multidisciplinary literature shows the appreciation of this phenomenon is widely conditioned by personal culture, disciplinary knowledge and professional know-how. From a research angle, analyzing complex systems from various points of view (user, technician, public authority, researcher) helps generate original development opportunities. This thesis results from a meeting between industrial engineering, political sciences and town planning around a logic of technology transfer. Indeed, the technologies of collaborative design and user-centric innovation are among the answers industrial engineering brings to problems generated by complexity. Our analysis brings us then to consider the potential that the adaptation of distributed collaborative design (DCD) represents. Consequently, we formulate tools, methods and one specific environment necessary for the development of a DCD dedicated to urban projects to facilitate their global and sustainable acceptability. These studies ask subsequently the question of role which should be given to the final user in the urban project design process. Political sciences help us to understand the relations between citizens and experts. This gives us elements to develop a collaborative approach from the conceptualization phase as this impacts the whole development of the projects. More globally, such collaborative processes may contribute to the emergence of sustainable cities ; La complexité des systèmes urbains est un défi constant pour les acteurs qui y sont confrontés. L'appréciation de ce phénomène est largement conditionnée par la posture (usager, technicien, pouvoir public, chercheur) et le bagage culturel, disciplinaire et professionnel, comme en rend compte une littérature pluridisciplinaire prolifique. Sur le plan de la recherche, multiplier les regards sur les systèmes complexes contribue à ...

National audience The automotive industry requires cost, quality and time control. To this end, PSA Peugeot Citroën has been working for a number of years on integrating professional knowledge into design support tools. More specifically, the aim is to produce and make available IT solutions to automate routine design tasks in the field of CAD. Following research and industrial production, the PSA Group is able to propose tools and methods that provide solutions to the needs of its users. Today, faced with a policy of globalisation of the market, the group puts forward three strands: growth, profitability and innovation. This innovation must now be reflected in the design process. The aim is therefore to identify and test methodological environments that are equipped to capitalise on consistent 'routine' and 'innovative' knowledge with a view to their automation. We will focus here on the development of such an environment which will be experienced in the field of door frames. ; National audience L'industrie automobile nécessite une maîtrise des coûts, de la qualité et des délais. Pour ce faire, PSA Peugeot Citroën travaille depuis quelques années sur l'intégration de connaissances métiers au sein d'outils d'aide à la conception. De manière plus précise, il s'agit de produire et de mettre à disposition des solutions informatiques permettant d'automatiser des tâches routinières de conception dans le domaine de la CAO. Suite à des travaux de recherche et des mises en productions industrielles, le Groupe PSA est en mesure de proposer des outils et des méthodes apportant des solutions aux besoins de ses utilisateurs. Aujourd'hui, face à une politique de globalisation du marché, le groupe met en avant trois axes : croissance, rentabilité et innovation. Cette innovation doit maintenant se retrouver dans le processus de conception. Il s'agit donc d'identifier et d'expérimenter des environnements méthodologiques outillés permettant de capitaliser les connaissances " routinières " et " innovantes " en cohérence, en vue ...

Machine designers must take into account different aspects of products such as technology, legislation and occupational safety. european directive 2006/42/ce promulgates machine safety design principles to prevent occupational risks. aimed at machine designers, its objective is to set out the need to obtain the lowest possible risk level according to the state of the art. although the designers of catalog machines have access to specific standards to perform a priori risk analyses, this is not the case for special machine designers. an original approach called ezid is presented in this report to help all machine designers in the task of hazard identification.based on the fact that hazards are linked to the presence of energies, ezid identifies hazards through the detection of design parameters linked to energy sources and flows. it then feeds back organized information to designers on every potential exchange of energy between the machine and its operators, bringing to light the need to add risk prevention solutions. ; Les concepteurs de machines de production doivent prendre en consideration les differents aspects des produits tels que la technologie, la legislation et la securite professionnelle. la directive europeenne 2006/42/ce definie les principes de prevention des risques professionnels. destinee aux concepteurs de machines, son objectif est d'imposer la necessite d'obtenir le niveau de risque le plus faible possible selon l'etat de l'art. bien que les concepteurs de machines catalogues ont acces a des normes specifiques pour realiser les analyses a priori, ce n'est pas le cas pour les concepteurs de machines speciales. une approche originale appelee ezid est presentee dans ce papier pour aider tous les concepteurs de machine pour l'identification des phenomenes dangereux.basee sur le fait que les phenomenes dangereux sont lies a la presence d'energie, ezid les identifie au travers de la detection des parametres de conception lies aux sources et flux d'energie. elle renvoie ensuite de maniere organisee ...

Machine designers must take into account different aspects of products such as technology, legislation and occupational safety. european directive 2006/42/ce promulgates machine safety design principles to prevent occupational risks. aimed at machine designers, its objective is to set out the need to obtain the lowest possible risk level according to the state of the art. although the designers of catalog machines have access to specific standards to perform a priori risk analyses, this is not the case for special machine designers. an original approach called ezid is presented in this report to help all machine designers in the task of hazard identification.based on the fact that hazards are linked to the presence of energies, ezid identifies hazards through the detection of design parameters linked to energy sources and flows. it then feeds back organized information to designers on every potential exchange of energy between the machine and its operators, bringing to light the need to add risk prevention solutions. ; Les concepteurs de machines de production doivent prendre en consideration les differents aspects des produits tels que la technologie, la legislation et la securite professionnelle. la directive europeenne 2006/42/ce definie les principes de prevention des risques professionnels. destinee aux concepteurs de machines, son objectif est d'imposer la necessite d'obtenir le niveau de risque le plus faible possible selon l'etat de l'art. bien que les concepteurs de machines catalogues ont acces a des normes specifiques pour realiser les analyses a priori, ce n'est pas le cas pour les concepteurs de machines speciales. une approche originale appelee ezid est presentee dans ce papier pour aider tous les concepteurs de machine pour l'identification des phenomenes dangereux.basee sur le fait que les phenomenes dangereux sont lies a la presence d'energie, ezid les identifie au travers de la detection des parametres de conception lies aux sources et flux d'energie. elle renvoie ensuite de maniere organisee ...

Radio communications using very broad frequency bands (UWB) have been developed rapidly since the early 2000s. This technique has advantages that made it attractive in many civil and military applications. The aim of this thesis is to design and characterize miniature Ultra Wide Band antennas in the time domain for the lower part of the frequency band (0.1-2 GHz). First, a new UWB antenna has been designed: the folded wire monopole antenna. This very simple and small size structure has naturally high bandwidth and quasi-omnidirectional radiation pattern over its entire band of operation. We develop an analytical model describing fast and accurate behavior of this antenna which enabled us to understand the role of different parameters of the antenna. There is excellent agreement between the results obtained by this model and those obtained by numerical simulation. With this analytical model, we can optimize the performance of the folded wire monopole for the low frequencies in two different ways: by thickening the diameter of radiating wires or by adding stubs. Both new antennas have been characterized and they have the same radiation characteristics, but a different matching. Typically, UWB antennas are characterized in the frequency domain (radiation pattern and antenna matching). However, a more adequate approach might be preferred: Ultra Wideband antennas can be described in the time-domain. We have introduced new descriptors which are used to quantify the distortion introduced by the antenna: fidelity function and time-domain gain. Time and frequency domain measurements have been performed. Measurements on the different versions of the folded wire antenna showed a good agreement with the simulations in terms of both frequency and time domain characterization. Finally, we have developed a very effective time domain technique measurement by using time domain reflectometry. This method gives very good results in terms of radiated pulses and antenna matching ; Les communications radio utilisant des très larges ...

Radio communications using very broad frequency bands (UWB) have been developed rapidly since the early 2000s. This technique has advantages that made it attractive in many civil and military applications. The aim of this thesis is to design and characterize miniature Ultra Wide Band antennas in the time domain for the lower part of the frequency band (0.1-2 GHz). First, a new UWB antenna has been designed: the folded wire monopole antenna. This very simple and small size structure has naturally high bandwidth and quasi-omnidirectional radiation pattern over its entire band of operation. We develop an analytical model describing fast and accurate behavior of this antenna which enabled us to understand the role of different parameters of the antenna. There is excellent agreement between the results obtained by this model and those obtained by numerical simulation. With this analytical model, we can optimize the performance of the folded wire monopole for the low frequencies in two different ways: by thickening the diameter of radiating wires or by adding stubs. Both new antennas have been characterized and they have the same radiation characteristics, but a different matching. Typically, UWB antennas are characterized in the frequency domain (radiation pattern and antenna matching). However, a more adequate approach might be preferred: Ultra Wideband antennas can be described in the time-domain. We have introduced new descriptors which are used to quantify the distortion introduced by the antenna: fidelity function and time-domain gain. Time and frequency domain measurements have been performed. Measurements on the different versions of the folded wire antenna showed a good agreement with the simulations in terms of both frequency and time domain characterization. Finally, we have developed a very effective time domain technique measurement by using time domain reflectometry. This method gives very good results in terms of radiated pulses and antenna matching ; Les communications radio utilisant des très larges bandes de fréquences (ULB) se sont rapidement développées depuis le début des années 2000. Cette technique présente des avantages qui l'ont rendue attractive dans un grand nombre d'applications civiles et militaires. L'objectif de cette thèse était de concevoir, de réaliser et de caractériser en régime impulsionnel des antennes Ultra Large Bande miniatures, dans la partie basse du spectre (0.1-2 GHz). Dans un premier temps, une nouvelle structure d'antenne ULB a été conçue: l'antenne filaire repliée. Cette structure très simple et de petite taille, possède naturellement une large bande passante ainsi qu'un diagramme de rayonnement quasi omnidirectionnel sur toute sa bande de fonctionnement. Nous avons développé un modèle analytique rapide et précis décrivant le comportement de cette antenne, ce qui nous a permis de comprendre le rôle des différents paramètres de l'antenne. Il y a un excellent accord entre les résultats obtenus par ce modèle et ceux obtenus par simulation numérique. Grâce à ce modèle analytique, nous avons pu optimiser les performances du monopole filaire replié vers les basses fréquences et ce, par deux méthodes : épaississement des brins rayonnants et ajout de tronçons de lignes. Les deux nouvelles antennes ainsi obtenues ont été caractérisées; elles présentent les mêmes caractéristiques de rayonnement, mais une adaptation différente. Les outils usuels de caractérisation des antennes dans le domaine fréquentiel (adaptation, diagramme de rayonnement) étant mal adaptés à une étude en mode impulsionnel, nous avons introduit de nouveaux descripteurs qui permettent de quantifier la distorsion introduite par l'antenne: Fonction de fidélité et gain temporel. Une campagne de mesures utilisant un banc impulsionnel et un banc de mesures fréquentiel a été réalisée. Les mesures effectuées sur les différentes versions de l'antenne filaire repliée ont montré un bon accord avec les simulations tant au niveau de la caractérisation fréquentielle que temporelle. Enfin, nous avons mis au point une technique de mesure temporelle très efficace en utilisant la réflectométrie temporelle. Cette méthode donne de très bons résultats en termes d'impulsions rayonnées et d'adaptation.

Radio communications using very broad frequency bands (UWB) have been developed rapidly since the early 2000s. This technique has advantages that made it attractive in many civil and military applications. The aim of this thesis is to design and characterize miniature Ultra Wide Band antennas in the time domain for the lower part of the frequency band (0.1-2 GHz). First, a new UWB antenna has been designed: the folded wire monopole antenna. This very simple and small size structure has naturally high bandwidth and quasi-omnidirectional radiation pattern over its entire band of operation. We develop an analytical model describing fast and accurate behavior of this antenna which enabled us to understand the role of different parameters of the antenna. There is excellent agreement between the results obtained by this model and those obtained by numerical simulation. With this analytical model, we can optimize the performance of the folded wire monopole for the low frequencies in two different ways: by thickening the diameter of radiating wires or by adding stubs. Both new antennas have been characterized and they have the same radiation characteristics, but a different matching. Typically, UWB antennas are characterized in the frequency domain (radiation pattern and antenna matching). However, a more adequate approach might be preferred: Ultra Wideband antennas can be described in the time-domain. We have introduced new descriptors which are used to quantify the distortion introduced by the antenna: fidelity function and time-domain gain. Time and frequency domain measurements have been performed. Measurements on the different versions of the folded wire antenna showed a good agreement with the simulations in terms of both frequency and time domain characterization. Finally, we have developed a very effective time domain technique measurement by using time domain reflectometry. This method gives very good results in terms of radiated pulses and antenna matching ; Les communications radio utilisant des très larges bandes de fréquences (ULB) se sont rapidement développées depuis le début des années 2000. Cette technique présente des avantages qui l'ont rendue attractive dans un grand nombre d'applications civiles et militaires. L'objectif de cette thèse était de concevoir, de réaliser et de caractériser en régime impulsionnel des antennes Ultra Large Bande miniatures, dans la partie basse du spectre (0.1-2 GHz). Dans un premier temps, une nouvelle structure d'antenne ULB a été conçue: l'antenne filaire repliée. Cette structure très simple et de petite taille, possède naturellement une large bande passante ainsi qu'un diagramme de rayonnement quasi omnidirectionnel sur toute sa bande de fonctionnement. Nous avons développé un modèle analytique rapide et précis décrivant le comportement de cette antenne, ce qui nous a permis de comprendre le rôle des différents paramètres de l'antenne. Il y a un excellent accord entre les résultats obtenus par ce modèle et ceux obtenus par simulation numérique. Grâce à ce modèle analytique, nous avons pu optimiser les performances du monopole filaire replié vers les basses fréquences et ce, par deux méthodes : épaississement des brins rayonnants et ajout de tronçons de lignes. Les deux nouvelles antennes ainsi obtenues ont été caractérisées; elles présentent les mêmes caractéristiques de rayonnement, mais une adaptation différente. Les outils usuels de caractérisation des antennes dans le domaine fréquentiel (adaptation, diagramme de rayonnement) étant mal adaptés à une étude en mode impulsionnel, nous avons introduit de nouveaux descripteurs qui permettent de quantifier la distorsion introduite par l'antenne: Fonction de fidélité et gain temporel. Une campagne de mesures utilisant un banc impulsionnel et un banc de mesures fréquentiel a été réalisée. Les mesures effectuées sur les différentes versions de l'antenne filaire repliée ont montré un bon accord avec les simulations tant au niveau de la caractérisation fréquentielle que temporelle. Enfin, nous avons mis au point une technique de mesure temporelle très efficace en utilisant la réflectométrie temporelle. Cette méthode donne de très bons résultats en termes d'impulsions rayonnées et d'adaptation.

System transition is required.to meet sustainability challenges. Our proposal is to apply general principles from a hybrid approach between forecasting (trends) and principled backcasting to industrial companies. These principles provide guidelines for each of the five dimensions of sustainability (political, territorial, social, environmental and economic). The developed method is based on the strategic process (analyze, choose, deploy) to include additional dimensions in the design of the product through the value constellation of the extended company. Two trajectories are explored: the integration of the environment (case studies) and the territory (theory) in product design through organizational innovation driven by the renewal of strategic and operational governance. This interdisciplinary research lays the foundation for the development of a method to support the transition of industrial companies towards the 5-dimensions sustainable development by integrating new dimensions hitherto little or no exploited in the product design process ; Pour répondre aux enjeux du développement durable, une transition systémique est nécessaire. Notre proposition est d'appliquer, au niveau de l'entreprise, des principes issus d'une approche hybride, entre forecasting (tendances) et principled backcasting. Ces principes proposent des lignes directrices pour chacune des cinq dimensions du développement durable (politique, territoire, Homme, environnement et économie). La méthode développée s'appuie sur le processus stratégique (analyse, choix et déploiement) pour intégrer des dimensions complémentaires dans la conception de produit à travers le réseau interne de création de valeur de l'entreprise étendue. Deux trajectoires ont étés explorées : l'intégration de l'environnement (cas d'études) et du territoire (théorie) dans la conception de produit grâce à l'innovation organisationnelle portée par le renouveau de la gouvernance stratégique et opérationnelle. Cette recherche interdisciplinaire pose les fondations pour le ...

System transition is required.to meet sustainability challenges. Our proposal is to apply general principles from a hybrid approach between forecasting (trends) and principled backcasting to industrial companies. These principles provide guidelines for each of the five dimensions of sustainability (political, territorial, social, environmental and economic). The developed method is based on the strategic process (analyze, choose, deploy) to include additional dimensions in the design of the product through the value constellation of the extended company. Two trajectories are explored: the integration of the environment (case studies) and the territory (theory) in product design through organizational innovation driven by the renewal of strategic and operational governance. This interdisciplinary research lays the foundation for the development of a method to support the transition of industrial companies towards the 5-dimensions sustainable development by integrating new dimensions hitherto little or no exploited in the product design process ; Pour répondre aux enjeux du développement durable, une transition systémique est nécessaire. Notre proposition est d'appliquer, au niveau de l'entreprise, des principes issus d'une approche hybride, entre forecasting (tendances) et principled backcasting. Ces principes proposent des lignes directrices pour chacune des cinq dimensions du développement durable (politique, territoire, Homme, environnement et économie). La méthode développée s'appuie sur le processus stratégique (analyse, choix et déploiement) pour intégrer des dimensions complémentaires dans la conception de produit à travers le réseau interne de création de valeur de l'entreprise étendue. Deux trajectoires ont étés explorées : l'intégration de l'environnement (cas d'études) et du territoire (théorie) dans la conception de produit grâce à l'innovation organisationnelle portée par le renouveau de la gouvernance stratégique et opérationnelle. Cette recherche interdisciplinaire pose les fondations pour le développement d'une méthode permettant de supporter la transition de l'entreprise industrielle vers le développement durable à 5 dimensions, par l'intégration dans le processus de conception de produit de dimensions jusqu'alors peu ou pas exploitées

The integration of various technologies, including computer and electronics, makes the nowadays designed systems increasingly complex. They have behaviors which are more elaborate and difficult to predict, they have a greater number of components in interaction and/or perform highest level functions. Parallel to this increasing complexity of these systems, the competitive of the global market imposes strong constraints of cost and time to the system developers. Other strong constraints deal with the quality of these systems, especially when they involve human risks or significant financial risks. Thus, developers are forced to adopt a rigorous design approach to meet the desired system requirements and satisfy the various constraints (cost, time, quality, dependability.). Several methodological approaches to guide the system design are defined through system engineering standards. Our work is based on the EIA-632 standard, which is widely used, especially in the aeronautical and military fields. It is to improve the systems engineering process described by the EIA-632, in order to incorporate a global and explicit consideration of dependability. Indeed, till now the dependability was achieved by reusing generic models after having studied and developed independently each function. So there was no specific consideration of the risks associated with the integration of several technologies. For this reason, we propose to concern ourselves with the dependability requirements at the global level and as early as possible in the development phase. Then, these requirements will be decline to lower levels. We based our approach on the processes of the EIA-632 standard that we expand. We also propose an original method for the declination of the dependability requirements based on fault trees and FMEAC, and an information model based on SysML in order to support our approach. An example from the aeronautical field illustrates our proposals. ; L'intégration de diverses technologies, notamment celles de l'informatique et ...

The integration of various technologies, including computer and electronics, makes the nowadays designed systems increasingly complex. They have behaviors which are more elaborate and difficult to predict, they have a greater number of components in interaction and/or perform highest level functions. Parallel to this increasing complexity of these systems, the competitive of the global market imposes strong constraints of cost and time to the system developers. Other strong constraints deal with the quality of these systems, especially when they involve human risks or significant financial risks. Thus, developers are forced to adopt a rigorous design approach to meet the desired system requirements and satisfy the various constraints (cost, time, quality, dependability.). Several methodological approaches to guide the system design are defined through system engineering standards. Our work is based on the EIA-632 standard, which is widely used, especially in the aeronautical and military fields. It is to improve the systems engineering process described by the EIA-632, in order to incorporate a global and explicit consideration of dependability. Indeed, till now the dependability was achieved by reusing generic models after having studied and developed independently each function. So there was no specific consideration of the risks associated with the integration of several technologies. For this reason, we propose to concern ourselves with the dependability requirements at the global level and as early as possible in the development phase. Then, these requirements will be decline to lower levels. We based our approach on the processes of the EIA-632 standard that we expand. We also propose an original method for the declination of the dependability requirements based on fault trees and FMEAC, and an information model based on SysML in order to support our approach. An example from the aeronautical field illustrates our proposals. ; L'intégration de diverses technologies, notamment celles de l'informatique et ...

The integration of various technologies, including computer and electronics, makes the nowadays designed systems increasingly complex. They have behaviors which are more elaborate and difficult to predict, they have a greater number of components in interaction and/or perform highest level functions. Parallel to this increasing complexity of these systems, the competitive of the global market imposes strong constraints of cost and time to the system developers. Other strong constraints deal with the quality of these systems, especially when they involve human risks or significant financial risks. Thus, developers are forced to adopt a rigorous design approach to meet the desired system requirements and satisfy the various constraints (cost, time, quality, dependability.). Several methodological approaches to guide the system design are defined through system engineering standards. Our work is based on the EIA-632 standard, which is widely used, especially in the aeronautical and military fields. It is to improve the systems engineering process described by the EIA-632, in order to incorporate a global and explicit consideration of dependability. Indeed, till now the dependability was achieved by reusing generic models after having studied and developed independently each function. So there was no specific consideration of the risks associated with the integration of several technologies. For this reason, we propose to concern ourselves with the dependability requirements at the global level and as early as possible in the development phase. Then, these requirements will be decline to lower levels. We based our approach on the processes of the EIA-632 standard that we expand. We also propose an original method for the declination of the dependability requirements based on fault trees and FMEAC, and an information model based on SysML in order to support our approach. An example from the aeronautical field illustrates our proposals. ; L'intégration de diverses technologies, notamment celles de l'informatique et ...

When modeling the mechanical behavior of structures, there often are uncertainties on rigidity or forces applied. Until now, safety coefficients are employed. This thesis takes place in the validation domain, which means validation of the numerical employed model with respect to reality. We try to bring practical tools for ingeniors to design a big structures, without over-sizing. This work uses the Lack-Of-Knowledge (LOK) theory, developed in LMT-Cachan in 2002. Its goal is to give an envelope response of the quantity of interest. We propose different extensions in order to modelize uncertainties on forces applied and on anisotropic rigidities. Another aspect of the work deals with the propagation of every kind of LOKs to a quasi-statical quantity of interest, like a local displacement or a local stress. We also develop a frame to obtain with such a model the probability-of-failure of the structure considered, and margin indicators which allow to evaluate the probability-of-failure improvement to result from better knowledge (and then reduced LOKs). Thank to these developments, the LOK theory enter in robust conception. ; Lors de la modélisation du comportement mécanique de structures comme des avions, ou des lanceurs spatiaux, il subsiste des incertitudes parfois fortes, notamment en ce qui concerne l'évaluation des efforts extérieurs ou la rigidité des liaisons. Jusqu'à présent, des coefficients de sécurité, souvent forfaitaires, sont employés, de manière à assurer le bonne tenue des structures, en se basant sur les résultats de modèles simplifiés de la réalité. Cette thèse s'inscrit dans la problématique de validation des modèles numériques utilisés, et tente d'apporter des outils pratiques pour le dimensionnement de toute structure réelle, en présence d'incertitudes, tout en limitant les sur-dimensionnements liés à l'éventuel emploi de coefficients de sécurité. Dans cette optique, les travaux présentés ici s'appuient sur la théorie des méconnaissances, qui vise à remplacer les politiques de marges par une ...