Le secteur aéronautique connaît un important essor depuis les années 1960, avec pour conséquence l’augmentation majeure des températures de fonctionnement des turbines. L’utilisation de capteurs, au plus proche de la chambre de combustion, est nécessaire pour maîtriser les performances des turboréacteurs. Cela justifie le besoin industriel de développer des assemblages céramique-métal résistants à un environnement sévère (>1100°C sous air, vibrations, etc.)L’alumine est sélectionnée en raison de son caractère isolant à haute température. La principale difficulté réside dans le choix du substrat métallique, qui doit être à la fois : réfractaire, résistant à l’oxydation, pour un coût abordable. Le potentiel d’un carbure ternaire (de type phase MAX) est évalué en tant que substrat métallique. La composition des brasures est ajustée pour chaque système afin de garantir : une tenue en température, une excellente ductilité et une compatibilité métallurgique vis-à-vis du substrat métallique.Des brasures ternaires Au-Pd-Pt sont formulées et élaborées. Les contours de solidus et de liquidus de ce ternaire sont déterminés expérimentalement. Les mécanismes de diffusion, de dissolution ou encore de pénétration inter-granulaire, à l’interface métal/brasure, sont décrits afin d’apporter des améliorations aux systèmes développés. Au cours du brasage, la diffusion de l’aluminium des substrats alumino-formeurs vers la brasure est prédominante, malgré la mise en place d’une barrière de diffusion. Le recours à des substrats métalliques nobles s’avère incontournable.Les paramètres géométriques et chimiques de l’assemblage sont établis pour chaque nouveau système étudié sur la base des processus physico-chimiques survenant au cours du brasage et du vieillissement sous air. Les meilleurs assemblages développés présentent une excellente herméticité après brasage. Les essais de vieillissement en cyclage thermique, dans des conditions sévères, sollicitent fortement la liaison céramique-brasure jusqu’à la rupture interfaciale. Deux voies d’améliorations sont proposées pour assurer la durabilité de la liaison céramique-brasure au cours du cyclage thermique. / Aerospace technology developments are blooming. Since the 1960’s the Turbine Entry Temperature for aero-engines gas turbines keeps rising to improve their efficiency. Sensors working close to the combustion chamber are required in order to master the turbine performances. This results in an industrial urge to develop ceramic-to-metal assemblies able to endure severe engine environment (>1100°C under air, vibrations, etc.)The choice of an alumina as the ceramic part ensures a good insulation at high temperature. The main issue remains the metallic material which must be refractory, resistant to oxidation and affordable. The application of a ternary carbide (phase MAX) is tested. The braze alloy composition is adjusted to each system so as to provide a good ductility, thermal stability and a metallurgical matching as regards dissolution and brittle compounds formation.Braze alloys based on the Au-Pd-Pt system are investigated. The aforesaid liquidus and solidus surfaces are defined from experimental measures. Diffusion, dissolution and inter-granular penetration at the metal/braze alloy interface are described to suggest improvements. Aluminum diffusion from alumina-forming materials towards the braze alloy is the dominant phenomenon during brazing, despite the use of a diffusion barrier. The use of noble materials for the metallic substrate is mandatory.Geometrical and chemical assembly parameters are defined for selected brazed system based on the physicochemical interactions occurring during brazing and aging under air. The best brazed assemblies present an excellent hermeticity after brazing. Thermal cycling aging in severe conditions is detrimental to ceramic-braze alloy bonding, leading to interfacial cracks. Two promising strategies are suggested to ensure a reliable ceramic-to-braze alloy bonding during thermal cycling.

La durée de vie utile résiduelle (RUL) peut être simplement définie comme une prédiction du temps restant pendant lequel un système est capable d'exécuter sa fonction prévue ; elle est mesurée à partir de l'instant présent jusqu'à la défaillance finale. Cette durée prévue dépend principalement de l'état de détérioration des composants du système et de leurs conditions de fonctionnement futures prévues. Ainsi, la prédiction de la RUL est un processus incertain et son contrôle n'est pas une tâche triviale. En général, le but de la prévision de la RUL est d'influencer la prise de décision pour le système. Dans cette thèse, on a présenté un cadre compréhensible pour le contrôle de la RUL. Les incertitudes du modèle ainsi que les perturbations du système ont été prises en compte dans le cadre proposé. Des questions telles que le traitement de l'incertitude et l'inclusion d'objectifs RUL dans la stratégie de contrôle sont étudiées, depuis la modélisation jusqu'à une architecture de contrôle globale finale. On a montré que l'on peut prédire la RUL à partir d'une estimation appropriée de la détérioration et d'hypothèses sur les conditions de fonctionnement futures. Les systèmes d'entraînement par friction sont utilisés pour illustrer l'utilité de l'architecture globale susmentionnée. Pour ce type de système, le frottement est à la fois source du mouvement et source de la détérioration. Ce double caractéristique de frottement est une motivation pour contrôler automatiquement la détérioration du système en maintenant un compromis entre les exigences de mouvement et les valeurs RUL souhaitées. Dans cette thèse, un nouveau modèle orienté contrôle pour les systèmes d'entraînement par friction, qui inclut un modèle dynamique de la détérioration, est proposé. Le degré de détérioration est considéré en fonction de l'énergie dissipée, à la surface de contact, pendant la transmission mécanique de puissance. Une approche est proposée pour estimer l'état actuel de la détérioration d'un système d'entraînement par friction. L'approche est basée sur un Filtre de Kalman Etendu (EKF en anglais) qui utilise un modèle augmenté incluant le système mécanique dynamique et la dynamique de détérioration. L'EKF fournit également des intervalles qui incluent sûrement la valeur de détérioration réelle avec une valeur de probabilité. Une nouvelle architecture de commande de la RUL est proposée, elle comprend : un système de surveillance de l'état de détérioration (par exemple l'EKF proposé), un estimateur de l'état de fonctionnement du système, un système de commande de la RUL et un principe actionneur de la RUL. L'estimateur des conditions de fonctionnement est basé sur l'hypothèse qu'il est possible de quantifier certaines caractéristiques des exigences de mouvement, par exemple le rapport cyclique des couples moteur. Le contrôleur RUL utilise une fonction de coût qui pondère les exigences de mouvement et les valeurs RUL souhaitées pour modifier un filtre à paramètres variables, utilisé ici comme principe actionneur RUL. Le principe actionneur RUL est basé sur une modification des couples exigés, provenant d'un éventuel système de contrôle de mouvement. Les résultats préliminaires montrent qu'il est possible de contrôler la RUL, selon le cadre théorique proposé. / Remaining Useful Lifetime (RUL) can be simply defined as a prediction of the remaining time that a system is able to perform its intended function, from the current time to the final failure. This predicted time mostly depends on the state of deterioration of the system components and their expected future operating conditions. Thus, the RUL prediction is an uncertain process and its control is not trivial task.In general, the purpose for predicting the RUL is to influence decision-making for the system. In this dissertation a comprehensive framework for controlling the RUL is presented. Model uncertainties as well as system disturbances have been considered into the proposed framework. Issues as uncertainty treatment and inclusion of RUL objectives in the control strategy are studied from the modeling until a final global control architecture. It is shown that the RUL can be predicted from a suitable estimation of the deterioration, and from hypothesis on the future operation conditions. Friction drive systems are used for illustrating the usefulness of the aforementioned global architecture. For this kind of system, the friction is the source of motion and at the same time the source of deterioration. This double characteristic of friction is a motivation for controlling automatically the deterioration of the system by keeping a trade-off, between motion requirements and desired RUL values. In this thesis, a new control-oriented model for friction drive systems, which includes a dynamical model of the deterioration is proposed. The amount of deterioration has been considered as a function of the dissipated energy, at the contact surface, during the mechanical power transmission. An approach to estimate the current deterioration condition of a friction drive system is proposed. The approach is based on an Extended Kalman Filter (EKF) which uses an augmented model including the mechanical dynamical system and the deterioration dynamics. At every time instant, the EKF also provides intervals which surely includes the actual deterioration value which a given probability. A new architecture for controlling the RUL is proposed, which includes: a deterioration condition monitoring system (for instance the proposed EKF), a system operation condition estimator, a RUL controller system, and a RUL actuation principle. The operation condition estimator is based on the assumption that it is possible quantify certain characteristics of the motion requirements, for instance the duty cycle of motor torques. The RUL controller uses a cost function that weights the motion requirements and the desired RUL values to modify a varying-parameter filter, used here as the RUL-actuating-principle. The RUL-actuating-principle is based on a modification of the demanded torques, coming from a possible motion controller system. Preliminary results show that it is possible to control de RUL according to the proposed theoretical framework.

La décrépitation (fracturation) des matériaux actifs de batteries associée à leur variation volumique lors des cycles de charge/décharge a pour effet d'accélérer leur corrosion par l'électrolyte et/ou d'induire une perte de connectivité électronique au sein de l'électrode ce qui réduit notablement leur durée de vie. C’est particulièrement le cas des hydrures métalliques utilisés dans les batteries Ni-MH. À ce jour, l'évaluation de leur fracturation se limite généralement à une observation post mortem des électrodes par microscopie ce qui ne permet pas une analyse détaillée du processus de décrépitation. À ce titre, un de nos principaux objectifs dans le cadre de ce travail de recherche a été de développer une méthode d'analyse novatrice et performante basée sur l'émission acoustique (EA) afin d'étudier in situ la fracturation d'électrodes négatives pour batteries Ni-MH. Dans une première étape, nous avons analysé en détail les signaux acoustiques produits lors de la charge (hydruration) d'un alliage commerciale à base de LaNi5 et d'un alliage MgNi obtenu par broyage mécanique. Nous avons ainsi pu séparer les signaux générés par la fracturation des particules d’hydrures métalliques de ceux associés à la formation de bulles de H2, ce qui a permis d’établir les mécanismes qui régissent leur fracturation. Par la suite, un montage expérimental, constitué d’une cellule électrochimique connectée à un capteur de force en compression et d’un équipement d’EA, a été mis point pour suivre in-situ la fracturation et la force générée par l’expansion/contraction lors du cyclage des électrodes MgNi et LaNi5. Nous avons ainsi pu confirmer que l’expansion/contraction volumique de l’alliage MgNi est plus progressif que pour l’alliage à base de LaNi5. Par la suite, l’étude comparée par EA des alliages MgNi, Mg0.9Ti0.1Ni et Mg0.9Ti0.1NiAl0.05 a permis de mettre en évidence l'influence de leur composition sur leur résistance à la pulvérisation. Finalement, nous avons étudié en détail l’influence de l’addition de palladium dans l’alliage Mg0.9Ti0.1NiAl0.05 sur son comportement électrochimique et sa résistance à la fracturation. / The pulverization (cracking) of active materials in batteries, induced by their volume change during charge/discharge cycles, accentuates their corrosion by the electrolyte and/or leads to a loss of electronic connectivity within the electrode, which notably reduces their cycle life. This particularly occurs for metallic hydrides used in Ni-MH batteries. To date, the evaluation of their cracking is generally limited to post mortem observations of the electrodes by microscopy, which does not allow for a detailed analysis of the decrepitation process. In this respect, one of our main research objectives was to develop an innovative and efficient analysis method based on acoustic emission (AE) for in situ monitoring of the cracking of negative electrodes for Ni-MH batteries. As a first step, a detailed analysis of the acoustic signals generated during the charge (hydriding) of a commercial LaNi5-based alloy and a MgNi alloy obtained by mechanical alloying was performed. This allowed separating the signals generated by the cracking of the metallic hydride particles from those induced by the formation of H2 bubbles. We have shown that the mechanism which governs the pulverization of the MgNi alloy remarkably differs from that of the LaNi5-based alloy. In a second step, an experimental set-up made of an electrochemical cell linked to a compression force cell and an AE equipment was elaborated, in order to monitor concomitantly the cracking and the force generated by the expansion/contraction of the MgNi and LaNi5 during cycling. We have thereby been able to confirm that the volume expansion/contraction of the MgNi alloy is more progressive than that of the LaNi5 alloy. The AE-based comparative study of MgNi, Mg0.9Ti0.1NiAl5 and Mg0.9Ti0.1NiAl0.05 alloys then allowed demonstrating the positive effect of the partial Mg substitution by Ti and adding of Al on the alloy decrepitation resistance. As a final step, we have studied the impact of palladium addition in the Mg0.9Ti0.1NiAl0.05 alloy on its electrochemical behaviour and cracking resistance.

The Astoria-Megler Bridge is a 6.6 kilometer (4.1 mile) long bridge, connecting Oregon and Washington on US 101, with a continuous steel truss main span of 376 m (1232 ft). It is the second longest main span bridge of this type in the world. Due to vortex shedding, some of the long truss verticals exhibit wind-induced torsional vibrations. These vibrations can create large numbers of repeated stress cycles in the truss verticals and the gusset plate assemblies. The members and connections were not designed for such conditions and the impact of this behavior on the service life of the bridge is uncertain. A full-scale representation of one of the truss verticals observed to exhibit such wind induced torsional response was fabricated and tested in the Structural Engineering Research Laboratory at Oregon State University. Experimental data of the rotational behavior and the stress distribution along the vertical were collected using inclinometers, an angular rate sensor, and uniaxial and rosette strain gages. The data collected were compared with existing analytical methods and predictions from finite element models. The observed experimental results including twist angle, stress distribution, and stress magnitude were well captured by both the finite element model and the analytical equations. Using analytical expressions, the fatigue lives of the existing bridge verticals were predicted based on assumed storm duration and recurrence. / Graduation date: 2013

As a consequence of the recent deregulation in the electrical power production industry, there has been a shift in the traditional ownership of power plants and the way they are operated. To hedge their business risks, the many new private entrepreneurs enter into long-term service agreement (LTSA) with third parties for their operation and maintenance activities. As the major LTSA providers, original equipment manufacturers have invested huge amounts of money to develop preventive maintenance strategies to minimize the occurrence of costly unplanned outages resulting from failures of the equipments covered under LTSA contracts. As a matter of fact, a recent study by the Electric Power Research Institute estimates the cost benefit of preventing a failure of a General Electric 7FA or 9FA technology compressor at $10 to $20 million. Therefore, in this dissertation, a two-phase data analytics approach is proposed to use the existing monitoring gas path and vibration sensors data to first develop a proactive strategy that systematically detects and validates catastrophic failure precursors so as to avoid the failure; and secondly to estimate the residual time to failure of the unhealthy items. For the first part of this work, the time-frequency technique of the wavelet packet transforms is used to de-noise the noisy sensor data. Next, the time-series signal of each sensor is decomposed to perform a multi-resolution analysis to extract its features. After that, the probabilistic principal component analysis is applied as a data fusion technique to reduce the number of the potentially correlated multi-sensors measurement into a few uncorrelated principal components. The last step of the failure precursor detection methodology, the anomaly detection decision, is in itself a multi-stage process. The obtained principal components from the data fusion step are first combined into a one-dimensional reconstructed signal representing the overall health assessment of the monitored systems. Then, two damage indicators of the reconstructed signal are defined and monitored for defect using a statistical process control approach. Finally, the Bayesian evaluation method for hypothesis testing is applied to a computed threshold to test for deviations from the healthy band. To model the residual time to failure, the anomaly severity index and the anomaly duration index are defined as defects characteristics. Two modeling techniques are investigated for the prognostication of the survival time after an anomaly is detected: the deterministic regression approach, and parametric approximation of the non-parametric Kaplan-Meier plot estimator. It is established that the deterministic regression provides poor prediction estimation. The non parametric survival data analysis technique of the Kaplan-Meier estimator provides the empirical survivor function of the data set comprised of both non-censored and right censored data. Though powerful because no a-priori predefined lifetime distribution is made, the Kaplan-Meier result lacks the flexibility to be transplanted to other units of a given fleet. The parametric analysis of survival data is performed with two popular failure analysis distributions: the exponential distribution and the Weibull distribution. The conclusion from the parametric analysis of the Kaplan-Meier plot is that the larger the data set, the more accurate is the prognostication ability of the residual time to failure model.

Das Örtliche Konzept dient der Berechnung von Lebensdauern zyklisch belasteter Konstruktionen. Dabei wird die Ermüdungsfestigkeit des ungekerbten Werkstoffes zugrundegelegt, um die Bauteillebensdauer abzuschätzen. Mit dieser Annahme verbunden ist ein starke Vereinfachung des Ermüdungsvorganges. Die Ursache ist die unzureichende Berücksichtigung wesentlicher Einflußgrößen, die die Festigkeit des interessierenden Bauteiles bestimmen und die andere Ausprägung als an der Werkstoffprobe erfahren. Dadurch ist die Zuverlässigkeit der Rechnung reduziert. Die vorliegende Arbeit untersucht und klassifiziert die Ursachen für dieses Übertragbarkeitsproblem. Mit einem sehr flexiblen multiplikativen Ansatz, der technologischen, statistischen und spannungsemchanischen Effekt berücksichtigt, läßt sich der Unterschied zwischen Festigkeit des Bauteils und der Werkstoffprobe rechnerisch erfassen. Zur numerischen Betrachtung technologischer und statistischer Einflüsse kann auf bereits bestehendes Wissen zurückgegriffen werden. Dies wird konzeptbezogen und mit Rücksicht auf die Anwendung angepaßt. Die Berücksichtigung des spannungsmechanischen Einflusses hingegen kann nicht von Bestehendem abgeleitet werden. Dazu wird aus einer umfangreichen Sammlung experimenteller Ergebnisse dieser Einflufaktor abgeleitet. Aus der so gegebenen Stichprobe lassen sich durch eine statistische Auswertung eine Reihe von Kennwerten identifizieren, die die Stützwirkung maßgeblich beeinflussen. Darauf aufbauend wird eine Bestimmungsgleichung für den spannungsmechanischen Einflußfaktor abgeleitet. Somit ist ein Stützwirkungskonzept geschaffen, das es erlaubt, die drei wesentliche Aspekte technologischen, statistischen und spannungsmechanischen Einfluß qualitativ uind quantitativ ins Örtliche Konzept zu integrieren. Mit dieser Vorgehsweise und einer vorliegenden Werkstoffwöhlerlinie läßt sich die bauteilspezifische Festigkeit abschätzen. Die so abgeleitete Wöhlerlinie erlaubt eine wesentlich genauere Lebensdauerberechnung, wie die Kotrolle an umfangreichen, unabhängigen Daten beweist. / The local strain concept serves to estimate the service life of cyclically loaded structures. Here the fatigue strength of the mere material coupon is taken as the basis in the calculation for notched components. This represents a distinct simplification of the complex fatigue process because of disregarding the relevant influences that determine strength and durability. By that the reliability of the calculation is not yet satisfying. The present investigation first classifies the reasons for this problem of transferabillity between notched component and unnotched specimen. A simple but flexible approach is used to describe the technological, statistical and and gradient effect. To consider the technological influences and statistical size effect numerically, known relationships and procedures are taken and adapted to the methodology of the concept. To catch gradient effects a new stress-relief-concept was developed. From a comprehensive database of experimental results gradient effects were were separated. By a statistical analysis relevant variables that govern the stress-relief are identified. Using these, an equation gets derived which enables to compute the gradient effects on fatigue strength. The developed stress-relief-concept allows to estimate a component-related strain(parameter-)-life-curve. Lifetime predictioins based on this are by far more reliable than those based on materials data only. This is verified by a check on an extensive and independent database.

Inadequate financial allocation for road maintenance is a threat to the impaired rural highways in Atlantic Canada. The conventional means of pavement rehabilitation has been to place a hot mix asphalt concrete overlay on the existing worn out pavement which is only a short term adjustment. The purpose is to provide a smooth wearing surface at a low cost. This traditional way of pavement repair does not fix the damage embedded within the pavement structure. After a certain extent of time the cracks in the original pavement start to reflect to the smooth new wearing surface, causing deterioration on the overlay. The advanced approach which is becoming more popular is the application of Full Depth Reclamation (FDR). This technique helps to repair the extensively defective roads by pulverizing the flexible pavement along with a fraction of the underlying damaged base layer. Thus a damage free base layer can be obtained by stabilizing and recompacting the pulverized materials. FDR is a sustainable and an environmentally beneficial repair method as it re-uses the in-situ materials. FDR process has been used around the world for over 25 years yet confronts some difficulties regarding the fluctuation in the strength of materials in various projects. It is inferred that some of these difficulties are due to the variability and poor quality in the restored materials. The variability in the recycled base layer is a result of currently utilizing a retroactive depth control method to attain a specific blend of asphalt concrete to granular base for the pulverized materials. Two FDR projects applying two different pulverization control methods (conventional retroactive and GPR depth control methods) were analyzed to investigate the improvements in consistency of the restored materials by using Ground Penetrating Radar (GPR). A wide range of asphalt concrete/base layer blend ratio was detected in retroactive control section, while consistent blend ratio was maintained in GPR survey by mapping the variability in the depth of pavement and sub-dividing the test sections accordingly. A GPR controlled constant blend ratio during pulverization displayed improvements in consistency of materials, physical and mechanical properties and performance as anticipated. The materials obtained by using the conventional retroactive depth control method exhibited higher variability in grain size distribution, optimum moisture content, optimum density, California Bearing Ratio, resilient modulus and shear strength. All materials from both projects exhibited excessive air voids and inadequate fines content as the as-obtained particles acted as conglomerated particles and enough fines were not generated after the pulverization. It is recommended that efficient quality control, precise specifications and appropriate pulverization methods will provide more reliable and impressive FDR pavements. / This thesis contains research on unstabilized full depth reclaimed aggregates properties

Modern aerospace systems rely heavily on legacy platforms and their derivatives. Historical examples show that after a vehicle design is frozen and delivered to a customer, successive upgrades are often made to fulfill changing requirements. Current practices of adapting to emerging needs with derivative designs, retrofits, and upgrades are often reactive and ad-hoc, resulting in performance and cost penalties. Recent DoD acquisition policies have addressed this problem by establishing a general paradigm for design for lifelong evolution. However, there is a need for a unified, practical design approach that considers the lifetime evolution of an aircraft concept by incorporating future requirements and technologies. This research proposes a systematic approach with which the decision makers can evaluate the value and risk of a new aircraft development program, including potential derivative development opportunities. The proposed Evaluation of Lifelong Vehicle Evolution (EvoLVE) method is a two- or multi-stage representation of the aircraft design process that accommodates initial development phases as well as follow-on phases. One of the key elements of this method is the Stochastic Programming with Recourse (SPR) technique, which accounts for uncertainties associated with future requirements. The remedial approach of SPR in its two distinctive problem-solving steps is well suited to aircraft design problems where derivatives, retrofits, and upgrades have been used to fix designs that were once but no longer optimal. The solution approach of SPR is complemented by the Risk-Averse Strategy Selection (RASS) technique to gauge risk associated with vehicle evolution options. In the absence of a full description of the random space, a scenario-based approach captures the randomness with a few probable scenarios and reveals implications of different future events. Last, an interactive framework for decision-making support allows simultaneous navigation of the current and future design space with a greater degree of freedom. A cantilevered beam design problem was set up and solved using the SPR technique to showcase its application to an engineering design setting. The full EvoLVE method was conducted on a notional multi-role fighter based on the F/A-18 Hornet.

Dans le contexte des systèmes commandés, l’effort de recherche est principalement porté sur la reconfiguration d’une loi de commande à l’apparition d’une situation de défaut. La reconfiguration a pour objectif de pallier au défaut et donc de maintenir les performances du système. La problématique principale de nos travaux est d’étudier ces systèmes du point de vue de leur sûreté de fonctionnement en s’interrogeant sur les causes qui engendrent une situation de défaut. Pour cela, il est supposé l’existence d’une relation étroite entre la commande d’un système, sa dégradation et ses défauts. Un cadre de modélisation stochastique de la dégradation est proposé intégrant l’usage du système ainsi que les différents modes de détérioration. Le pronostic de la durée de vie résiduelle RUL de l’actionneur -élément critique de ces systèmes- est dérivé de l’ensemble des modèles. La RUL est alors utilisée comme un outil de reconfiguration de la loi LQR (Linear Quadratic Regulator) d’un système mono-actionné dans le cadre d’une maintenance prédictive. L’impact de cette nouvelle politique de maintenance sur les performances statiques et dynamiques du système est évalué. Enfin, le comportement stochastique d’un système tolérant aux fautes basé sur une redondance d’actionneurs est étudié au travers des modèles développés / In the context of control systems, the research effort is focused on how to reconfigure the control law upon the occurrence of a faulty situation. The reconfiguration procedure aims to overcome the fault and thus to maintain system performances. The main issue of this thesis is to study these systems in terms of their dependability by questioning the causes that generate a fault. Then it is assumed a close relationship between the control of a system, its degradation and its faults. A stochastic modelling framework is proposed combining the use of the system and the various modes of deterioration. The actuator is assumed to be the most critical part of a system. The prognosis of its remaining life RUL is derived from these models. This RUL is then used as a tool for reconfiguring the LQR law (Linear Quadratic Regulator) of a system with a single actuator in the context of a predictive maintenance. The impact of this new maintenance policy on static and dynamic performances is assessed. Finally the stochastic behavior of a fault tolerant control system is studied by means of the achieved models

Cette thèse est consacrée à la description, la prédiction et la prévention des défaillances de systèmes. Elle se compose de quatre parties relatives à la modélisation stochastique de dégradation, au pronostic de défaillance du système, à l'estimation du niveau de défaillance et à l'optimisation de maintenance.Le processus d'Ornstein-Uhlenbeck (OU) dépendant du temps est introduit dans un objectif de modélisation des dégradations. Sur la base de ce processus, le premier instant de passage d’un niveau de défaillance prédéfini est considéré comme l’instant de défaillance du système considéré. Différentes méthodes sont ensuite proposées pour réaliser le pronostic de défaillance. Dans la suite, le niveau de défaillance associé au processus de dégradation est estimé à partir de la distribution de durée de vie en résolvant un problème inverse de premier passage. Cette approche permet d’associer les enregistrements de défaillance et le suivi de dégradation pour améliorer la qualité du pronostic posé comme un problème de premier passage. Le pronostic de défaillances du système permet d'optimiser sa maintenance. Le cas d'un système contrôlé en permanence est considéré. La caractérisation de l’instant de premier passage permet une rationalisation de la prise de décision de maintenance préventive. L’aide à la décision se fait par la recherche d'un niveau virtuel de défaillance dont le calcul est optimisé en fonction de critères proposés / This thesis is dedicated to describe, predict and prevent system failures. It consists of four issues: i) stochastic degradation modeling, ii) prognosis of system failures, iii) failure level estimation and iv) maintenance optimization. The time-dependent Ornstein-Uhlenbeck (OU) process is introduced for degradation modeling. The time-dependent OU process is interesting from its statistical properties on controllable mean, variance and correlation. Based on such a process, the first passage time is considered as the system failure time to a pre-set failure level. Different methods are then proposed for the prognosis of system failures, which can be classified into three categories: analytical approximations, numerical algorithms and Monte-Carlo simulation methods. Moreover, the failure level is estimated from the lifetime distribution by solving inverse first passage problems. This is to make up the potential gap between failure and degradation records to reinforce the prognosis process via first passage problems. From the prognosis of system failures, the maintenance optimization for a continuously monitored system is performed. By introducing first passage problems, the arrangement of preventive maintenance is simplified. The maintenance decision rule is based on a virtual failure level, which is solution of an optimization problem for proposed objective functions