WorldCat Identities

Fouladirad, Mitra (1977-....).

Works: 24 works in 26 publications in 2 languages and 26 library holdings
Roles: Other, Opponent, Thesis advisor, Author
Publication Timeline
Most widely held works by Mitra Fouladirad
Contribution au pronostic d'une pile à combustible de type PEMFC : approche par filtrage particulaire by Marine Jouin( )

2 editions published in 2015 in French and held by 2 WorldCat member libraries worldwide

The development of new energy converters, more efficient and environment friendly, such as fuelcells, tends to accelerate. Nevertheless, their large scale diffusion supposes some guaranties in termsof safety and availability. A possible solution to do so is to develop Prognostics and HealthManagement (PHM) on these systems, in order to monitor and anticipate the failures, and torecommend the necessary actions to extend their lifetime. In this spirit, this thesis deals with theproposal of a prognostics approach based on particle filtering dedicated to PEMFCs.The reasoning focuses first on setting a formalization of the working framework and theexpectations. This is pursued by the development of a physic-based modelling enabling a state ofhealth estimation and its evolution in time. The state estimation is made thanks to particle filtering.Different variants of filters are considered on the basis of the literature and new proposals adaptedto PHM are proposed and compared to existing ones. State of health estimates given by the filter areused to predict the future state of the system and its remaining useful life. All the proposals arevalidated on four datasets from PEMFC following different mission profiles. The results show goodperformances for predictions and remaining useful life estimates before failure
Analyse statistique de processus stochastiques : application sur des données d'orages by Van-Cuong Do( )

1 edition published in 2019 in English and held by 2 WorldCat member libraries worldwide

The work presented in this PhD dissertation concerns the statistical analysis of some particular cases of the Cox process. In a first part, we study the power-law process (PLP). Since the literature for the PLP is abundant, we suggest a state-of-art for the process. We consider the classical approach and recall some important properties of the maximum likelihood estimators. Then we investigate a Bayesian approach with noninformative priors and conjugate priors considering different parametrizations and scenarios of prior guesses. That leads us to define a family of distributions that we name H-B distribution as the natural conjugate priors for the PLP. Bayesian analysis with the conjugate priors are conducted via a simulation study and an application on real data. In a second part, we study the exponential-law process (ELP). We review the maximum likelihood techniques. For Bayesian analysis of the ELP, we define conjugate priors: the modified- Gumbel distribution and Gamma-modified-Gumbel distribution. We conduct a simulation study to compare maximum likelihood estimates and Bayesian estimates. In the third part, we investigate self-exciting point processes and we integrate a power-law covariate model to this intensity of this process. A maximum likelihood procedure for the model is proposed and the Bayesian approach is suggested. Lastly, we present an application on thunderstorm data collected in two French regions. We consider a strategy to define a thunderstorm as a temporal process associated with the charges in a particular location. Some selected thunderstorms are analyzed. We propose a reduced maximum likelihood procedure to estimate the parameters of the Hawkes process. Then we fit some thunderstorms to the power-law covariate self-exciting point process taking into account the associated charges. In conclusion, we give some perspectives for further work
Contribution to prognostics of proton exchange membrane fuel cells : approaches based on degradation information at multiple levels by Dacheng Zhang( )

1 edition published in 2018 in English and held by 2 WorldCat member libraries worldwide

In the context of the energy transition, fuel cell becomes one of the promising alternative energy sources. Recently the spotlight is on fuel cell systems research, and more particularly on Proton Exchange Membrane Fuel Cell (PEMFCs) which is one of the best candidates for both stationary and transportation applications. Even if this technology is close to being competitive, it is not yet ready to be considered for a large scale industrial deployment because of its limited durability and reliability. Prognostics and Health Management (PHM) is a recent approach to manage and possibly extend life duration of technological systems. Prognostic techniques can provide an estimation of fuel cell State Of Health (SOH) and a prediction for their Remaining Useful Life (RUL) to help the manufacturers improving fuel cell performance and managing its lifespan.The objective of this work is to develop prognostic methodologies for the RUL prognosis adapted to the complexity of PEMFCs. Indeed, the PEMFC is a multi-scale and multi-physics system, and various challenges are faced:1. The definition of SOH to build a degradation indicator.2. The coexistence of both reversible and irreversible degradation phenomena.3. Taking into account different deterioration causes and effects of operating conditions.In the first part of our work, we conduct a state of the art analysis on PHM for PEMFCs, with the aim of proposing a SOH definition and building a degradation indicator for PEMFC prognosis purposes. And since PEMFC measurements are scarce, the state of the art on Lithium batteries, other electrochemical cells, is also explored.In the second part, we develop a particle filtering based prognostic algorithm for PEMFC, based on output power measurements. The first results show that the prognosis algorithm is disturbed by the existing reversible degradation. However, the irreversible degradation can be estimated thanks to characterization tests, such as Electrochemical Impedance Spectroscopy (EIS), which is applied from time to time. We propose thus an adapted & extended prognostic algorithm to take into account both health indicators: the output power degradation and the SOH degradation estimated from EIS characterization. The performance of the proposed algorithm is evaluated by different prognostic performance metrics, and the results show the interest of this approach.In the third part, the problem is addressed from a more theoretical point of view. Indeed, a system's degradation behavior is often correlated with internal and external covariates which are usually difficult to access owing to expensive measurement cost. Therefore, we first developed a prognostic approach with online inspections on the degradation covariate at a different level, and then we propose an approach for RUL prognosis based on an ensemble of models using different sources at different levels. The RUL predictions of both models are dynamically aggregated on the basis of prognostic performance evaluated on a set of historical data. Consequently, the prediction accuracy is improved by overcoming both models' drawbacks and leveraging their strengths. In the last part, we extend the problem to multi-level prognostics and explore new possibilities, which open new aspects for future research on PEMFC lifetime prognosis and management
Amélioration de la chaine logistique de pièces de rechange en boucle fermée : application des modèles d'apprentissage by Hamza El Garrab( )

1 edition published in 2020 in French and held by 2 WorldCat member libraries worldwide

In the field of after-sales service and particularly in maintenance, the quick intervention and repair of the customer's property is a key element for his satisfaction and for the creation of the brand image in the market. The work presented in this thesis proposes a Big Data and Machine Learning approach for the improvement of the information flow in the spare parts supply chain. Our contribution focuses on load forecasting in spare parts repair centers, which are the main suppliers of parts used to repair customers' systems. The size of the supply chain and its complexity, the large number of part numbers as well as the multitude of special cases (countries with specific laws, special parts...) makes that classical approaches do not offer reliable forecasts for repair services. In this project, we propose learning algorithms allowing the construction of knowledge from large volumes of data, instead of manual implementation. We will see the models in the literature, present our methodology, and then implement the models and evaluate their performance in comparison with existing algorithms
Surveillance préventive des roulements par analyse multi-capteurs by Guillaume Bruand( )

1 edition published in 2019 in English and held by 2 WorldCat member libraries worldwide

La surveillance préventive est une approche courante permettant de réduire les coûts associés à la maintenance en milieu industriel. En effet, un diagnostic précoce peut prévenir des dommages critiques sur une machine donnée, et permet à l'utilisateur de planifier la maintenance afin de minimiser le temps d'immobilisation du moyen de production. Dans cette thèse il est montré que les capteurs d'angle sont particulièrement adaptés au diagnostic des machines tournantes, et plus spécifiquement à la détection des défauts de roulement. Ils sont combinés de manière avantageuse afin d'étudier l'orbite d'un arbre tournant, celle-ci donnant des informations pertinentes sur l'état de fonctionnement de la machine. Un modèle mécanique original est proposé afin de décrire les déplacements d'un arbre en rotation en présence d'un roulement défectueux. Des caractéristiques sont extraites de l'orbite étudiée, et utilisées comme un indicateur de sévérité de défaillance. Les résultats montrent que la méthode proposée a de multiples avantages sur les approches plus conventionnelles, telles que celles basées sur les accéléromètres, et représente ainsi une alternative intéressante dans un contexte industriel
System Reliability : Inference for Common Cause Failure Model in Contexts of Missing Information by Huu Du Nguyen( )

1 edition published in 2019 in English and held by 2 WorldCat member libraries worldwide

The effective operation of an entire industrial system is sometimes strongly dependent on the reliability of its components. A failure of one of these components can lead to the failure of the system with consequences that can be catastrophic, especially in the nuclear industry or in the aeronautics industry. To reduce this risk of catastrophic failures, a redundancy policy, consisting in duplicating the sensitive components in the system, is often applied. When one of these components fails, another will take over and the normal operation of the system can be maintained. However, some situations that lead to simultaneous failures of components in the system could be observed. They are called common cause failure (CCF). Analyzing, modeling, and predicting this type of failure event are therefore an important issue and are the subject of the work presented in this thesis. We investigate several methods to deal with the statistical analysis of CCF events. Different algorithms to estimate the parameters of the models and to make predictive inference based on various type of missing data are proposed. We treat confounded data using a BFR (Binomial Failure Rare) model. An EM algorithm is developed to obtain the maximum likelihood estimates (MLE) for the parameters of the model. We introduce the modified-Beta distribution to develop a Bayesian approach. The alpha-factors model is considered to analyze uncertainties in CCF. We suggest a new formalism to describe uncertainty and consider Dirichlet distributions (nested, grouped) to make a Bayesian analysis. Recording of CCF cause data leads to incomplete contingency table. For a Bayesian analysis of this type of tables, we propose an algorithm relying on inverse Bayes formula (IBF) and Metropolis-Hasting algorithm. We compare our results with those obtained with the alpha- decomposition method, a recent method proposed in the literature. Prediction of catastrophic event is addressed and mapping strategies are described to suggest upper bounds of prediction intervals with pivotal method and Bayesian techniques. Recent events have highlighted the importance of reliability redundant systems and we hope that our work will contribute to a better understanding and prediction of the risks of major CCF events
Modélisation probabiliste du pronostic : application à un cas d'étude et à la prise de décision en maintenance by Khanh Le Son( Book )

in French and held by 1 WorldCat member library worldwide

Remaining useful life (RUL) estimation is a major scientific challenge and a principal topic in the scientific community which takes an interest to prognosis problems. The use of tools and methods collected under the terms of prognostic is widely developed in many domains as aerospace industry, electronics, medicine, etc. The common underlying problem is the implementation of models which can take into account on-line the data histories of system and its environment, the diagnosis on its current state and possibly the future operational conditions for predicting the residual lifetime. In this context, the principal problem of our works is the use of probabilistic approaches (type of non-stationary stochastic process) to construct the innovatory prognostic models from a degradation indicator of system and to use the residual lifetime prediction for maintenance implementation. The advantage of these models is to have the regularity proprieties which make easy the probability calculation and RUL estimation. In order to test the performances of our models, a comparative study is carried out on the data provided by the 2008 IEEE Prognostic and Health Management (PHM)
Estimation précise et robuste de l'état de vieillissement de piles PEMFC par observateurs bayésiens dans le cadre d'une approche basée modèle by Andres Jacome( )

1 edition published in 2021 in French and held by 1 WorldCat member library worldwide

PHM (Prognostics and Health Management) represents a real opportunity to improve fuel cell performance and extend the life of fuel cells. This field of study has recently gained much interest. The main goal is to make optimum use of the data measured by all available sensors in order to evaluate the specific indicators of PEMFC ageing and possibly modify the operation of the fuel cell in order to optimize its lifetime. The proposed PhD is part of a model-based approach and willbe based on the expertise in fuel cell modelling developed at the Modelling Laboratory. An on-line estimator of the ageing state of the fuel cell will be developed. The proposed observer presents the characteristic of combining a state model derived from the MePHYSTO fuel cell model with the different data sensors available (voltage, current, pressure, temperature). The envisaged method makes it possible to jointly estimate the state variables, and in particular the ageing state, as well as to update the model parameters. Given the nature of the state variables to be estimated, we will move towards sophisticated observers adapted to non-linear and non-Gaussian problems in order to obtain a solution approaching the optimal Bayesian estimate
Inférence statistique et équations différentielles stochastiques. Applications en hydrologie. by Jasmine Cesars( )

1 edition published in 2019 in French and held by 1 WorldCat member library worldwide

Stochastic differential equations (SDE) are often used to model random phenomenain continuous time. This is the case for SDE whose solution are diffusion processesdescribing propagations of diseases or financial stocks. The study of SDE governed bya Wiener process (or Brownian motion) has made good progress in recent years, butthe SDE governed by Levy processes jump are less studied due to their complexity.In this PhD work, we are interested in SDE with jumps, having an explicit solutionsuch as the Black-Scholes model governed by a Poisson process associated withstochastic jumps. The Langevin process with random jumps is also studied. Thedistributional properties of these models are presented, in particular the fact thatthe direct or transformed solutions of the associated SDE can be processes withindependent increments. The link with the probabilistic characteristics of the jumpamplitudes is highlighted. In practice, the observation of a solution process of theseSDE can be carried out only in discrete time whereas it is a continuous time process.The results, which we have obtained concerning the laws of probability associatedwith discrete time observations, allow to establish conditional likelihood useful forstatistical inference on the model parameters. Thus, the study of the logarithm of thelikelihood ratio is conducted in the case of the Black-Scholes model with jumps andchange points. A change point test about the intrinsic rate of decrease is proposedas well as methods of numerical simulations of the SDE solutions. Scripts writtenin the programming environment allows to generate artificial data sets offeringpossibilities to test inferential tools. An application in hydrology is carried out fromdata concerning Guadeloupe and from the HYDRO bank
Détection de défaillance dans un système stochastique linéaire en présence de paramètres de nuisance by Mitra Fouladirad( Book )

2 editions published in 2005 in French and held by 1 WorldCat member library worldwide

In chapter 1, we recall some basic results of the theory of statistical hypotheses testing and we present some statistical tools that will be used in the following chapters. In chapter 2, the problem of fault detection in linear stochastical systems is solved by using the theory invariant tests and UBCP tests of Wald. We propose a UBCP test for fault detection in linear models with nuisance parameters. We analyse the impact of the different nuisance elimination methods on the statistical properties of the UBCP test. These results are applied to a dynamical system where the nuisance parameters are presented by a state equation. We conclude that the use of the state equation improves the statistical properties of the UBCP test. Chapter 3 is devoted to the on line change detection in a linear model with nuisance parameters. We analyse the properties of the nuisance rejection matrix and we propose a change detection algorithm of type GLR. The statistical properties of this algorithm are exposed. These properties depend essentially on the size of the sliding window. In chapter 4 some examples of the GPS integrity monitoring illustrate the theoretical results of chapter 2 by using numerical simulation
Diagnostic des équipements de production de semi-conducteurs par analyse statistique by Julien Marino( )

1 edition published in 2017 in French and held by 1 WorldCat member library worldwide

This thesis presents the construction of a fault detection algorithm, which can be used in real-time for semiconductor processes. The methodology is based on the statistical analysis of the data collected on the equipments. Faults are detected both with supervised learning, relatively to past data of a given equipment, and with unsupervised learning, relatively to the present data of groups of equipments. The proposed approach is tested on real equipments from STMicroelectronics, using faults which effectively occurred in the industrial environment.Each equipment and each of its operating modes is individually modeled, using reference data. A statistical test, based on such model, is performed every time a process is realized. This test is named Gaussian Time Error, because it is based on Gaussian models computed at each measurement time. By identifying differences in the new data relatively to the model, it allows to detect faults on the equipments.In the semiconductor industry, equipements are frequently maintained, which modifies the observed data. As a result, we introduce a specific test for the maintenances, to verify that no human error was made during the intervention. When no fault is seen, several parameters of the Gaussian Time Error are updated. This procedure is performed on a small quantity of data, thus allowing the equipment to be monitored quickly after maintenance.On top of the individual monitoring of equipments, we propose to gather the data of all equipments performing identical processes, in order to provide an unsupervised fault detection method
Pronostic de la performance d'Efficacité Energétique pour la prise de décision en maintenance dans les systèmes industriels by Anh Hoang( )

1 edition published in 2017 in English and held by 1 WorldCat member library worldwide

Among sustainability consideration, energy is today the key for economic growth in industrial systems. Energy resources are however limited and becomes more and more expensive. The energy optimization of manufacturing systems must therefore be considered as a major challenge to be compliant with environmental impact and management of energy resources. This should be reflected primarily by using energy efficiency (EE) as main key lever to deploy sustainability to plants, i.e. reduce the amount of energy required to provide products and services. With regards to this EE context, the aim of this thesis is to investigate the problem of considering energy efficiency and its prediction as a new indicator in maintenance decision-making. In that way, we develop first a concept of energy efficiency, called EEI (energy efficiency indicator), applicable to the different levels of abstraction of an industrial system. Then, we propose a generic formulation to evaluate the EEI (and its evolution) taking into account static and dynamic factors of influence. The temporal evolution of this indicator with respect to the degradation of the system is addressed in a predictive maintenance objective. It leads to found an energy efficiency performance concept called REEL (remaining energy-efficient lifetime), representing the residual energy lifetime. To predict the potential evolution of the IEE to calculate REEL, a generic approach based on existing predictive approaches is also developed. Next, we investigate the use of EE in CBM maintenance decision-making. Finally, all these contributions are validated on the TELMA platform
Contribution à l'optimisation des politiques de maintenance et l'analyse de risque dans la planification des opérations d'assemblage - désassemblage à deux niveaux by Zouhour Guiras( )

1 edition published in 2019 in French and held by 1 WorldCat member library worldwide

La réalité des marchés économiques impose des contraintes aux entreprises manufacturières qui sont de plus en plus difficiles à réaliser, comme la diversification des produits, l'amélioration de leur qualité, la réduction des coûts et la diminution des retards. Ces contraintes sont satisfaites par une meilleure organisation des systèmes de fabrication en utilisant les ressources techniques existantes. Notre présente thèse met l'accent sur deux contributions majeures, la première consiste à modéliser différents cas du système industriel (Système de production simple, système d'assemblage, système de désassemblage) en intégrant des politiques de maintenance adéquates. La deuxième contribution repose sur l'évaluation des risques de pertes de profit d'une décision prise suite à l'optimisation des différents systèmes industriels étudiés. Trois différents problèmes industriels sont étudiés, le premier concerne le développement des méthodes d'évaluation de risque de perte de profit résultant du choix d'un algorithme d'optimisation pour résoudre un problème de planification conjointe de production et de maintenance. Pour atteindre nos objectifs, nous commençons par calculer les plans de production et de maintenance en utilisant différents algorithmes d'optimisation. En outre, nous proposons des modèles analytiques pour quantifier le risque de perte de profit résultant des retours de produits et de la prise en compte des durées de réparation de pannes non nulles. Cette étude fournit des informations sur les algorithmes d'optimisations les plus efficaces pour les problématiques rencontrés pour aider et orienter les décideurs dans l'analyse et l'évaluation de leurs décisions. La deuxième problématique concerne l'optimisation de la planification du système d'assemblage à deux niveaux. Un modèle mathématique est développé pour incorporer une planification de l'approvisionnement pour les systèmes d'assemblage à deux niveaux dont les délais d'approvisionnement et les pannes du système sont stochastiques. La planification de maintenance optimale obtenue est utilisée dans l'évaluation des risques afin de trouver la période seuil de réparation qui réduit les pertes de profit. La troisième problématique étudiée concerne l'optimisation de la planification dans le cadre d'assemblage à base de désassemblage des produits usagés en tenant compte de la dégradation du système de production. Un modèle analytique est développé pour envisager le désassemblage, la remise à neuf des produits usagés qui contribuent à l'assemblage des produits finis. En effet, ces derniers peuvent être constitués de composants neufs ou remis à neuf. Une politique de maintenance est séquentiellement intégrée pour réduire l'indisponibilité du système. Le but de cette étude est d'aider les décideurs, dans certaines conditions, à choisir le processus le plus rentable pour satisfaire le client et qui peut également s'adapter aux risques potentiels qui peuvent perturber le système de désassemblage-assemblage. Le risque lié aux périodes de réparation du système est discuté, ce qui a un impact sur la prise de décision managériale
Analyse et optimisation de la fiabilité d'un équipement opto-électrique équipé de HUMS by Camille Baysse( )

1 edition published in 2013 in French and held by 1 WorldCat member library worldwide

Dans le cadre de l'optimisation de la fiabilité, Thales Optronique intègre désormais dans ses équipements, des systèmes d'observation de leur état de fonctionnement. Cette fonction est réalisée par des HUMS (Health & Usage Monitoring System). L'objectif de cette thèse est de mettre en place dans le HUMS, un programme capable d'évaluer l'état du système, de détecter les dérives de fonctionnement, d'optimiser les opérations de maintenance et d'évaluer les risques d'échec d'une mission, en combinant les procédés de traitement des données opérationnelles (collectées sur chaque appareil grâce au HUMS) et prévisionnelles (issues des analyses de fiabilité et des coûts de maintenance, de réparation et d'immobilisation). Trois algorithmes ont été développés. Le premier, basé sur un modèle de chaînes de Markov cachées, permet à partir de données opérationnelles, d'estimer à chaque instant l'état du système, et ainsi, de détecter un mode de fonctionnement dégradé de l'équipement (diagnostic). Le deuxième algorithme permet de proposer une stratégie de maintenance optimale et dynamique. Il consiste à rechercher le meilleur instant pour réaliser une maintenance, en fonction de l'état estimé de l'équipement. Cet algorithme s'appuie sur une modélisation du système, par un processus Markovien déterministe par morceaux (noté PDMP) et sur l'utilisation du principe d'arrêt optimal. La date de maintenance est déterminée à partir des données opérationnelles, prévisionnelles et de l'état estimé du système (pronostic). Quant au troisième algorithme, il consiste à déterminer un risque d'échec de mission et permet de comparer les risques encourus suivant la politique de maintenance choisie.Ce travail de recherche, développé à partir d'outils sophistiqués de probabilités théoriques et numériques, a permis de définir un protocole de maintenance conditionnelle à l'état estimé du système, afin d'améliorer la stratégie de maintenance, la disponibilité des équipements au meilleur coût, la satisfaction des clients et de réduire les coûts d'exploitation
Modèles stochastiques pour l'évaluation de politiques de maintenance sur des systèmes à dégradation graduelle by Amélie Ponchet-Durupt( Book )

in French and held by 1 WorldCat member library worldwide

Most of systems in the industry are gradually deteriorating systems or devices which undergo a change in their deterioration rate, e.g., be-cause of environmental or use conditions. This change can impact the degradation of the sys-tem and leads to a sudden increase of the system's deterioration rate. The consideration of such changes leads to a more realistic degradation model and opens perspectives in a maintenance decision making point of view. Several condition-based maintenance policies are then suggested and each of them is adapted to the available, or used, information on the system. The impact of the available information on the average long-run cost rate of the maintained system is studied for each maintenance policy considering perfect maintenance actions. In practical cases, maintenance actions do not lead to the renewal of the system. Hence a second study is performed on a maintenance policy adapted to large structures such as bridges, dikes, or pipelines. Such systems are built in order to be operational for a given finite time span which can be seen, for example, as an insurance deadline. Based on existing imperfect maintenance models, several improvement functions which model the impact of an imperfect maintenance action on the system, are proposed. A systematic maintenance policy is then considered and evaluated on a finite time span
A holistic framework of degradation modeling for reliability analysis and maintenance optimization of nuclear safety systems by Yanhui Lin( )

1 edition published in 2016 in French and held by 1 WorldCat member library worldwide

Composants de systèmes de sûreté nucléaire sont en général très fiable, ce qui conduit à une difficulté de modéliser leurs comportements de dégradation et d'échec en raison de la quantité limitée de données disponibles. Par ailleurs, la complexité de cette tâche de modélisation est augmentée par le fait que ces systèmes sont souvent l'objet de multiples processus concurrents de dégradation et que ceux-ci peut être dépendants dans certaines circonstances, et influencé par un certain nombre de facteurs externes (par exemple la température, le stress, les chocs mécaniques, etc.).Dans ce cadre de problème compliqué, ce travail de thèse vise à développer un cadre holistique de modèles et de méthodes de calcul pour l'analyse basée sur la fiabilité et la maintenance d'optimisation des systèmes de sûreté nucléaire en tenant compte des connaissances disponibles sur les systèmes, les comportements de dégradation et de défaillance, de leurs dépendances, les facteurs influençant externes et les incertitudes associées.Les contributions scientifiques originales dans la thèse sont:(1) Pour les composants simples, nous intégrons des chocs aléatoires dans les modèles de physique multi-états pour l'analyse de la fiabilité des composants qui envisagent dépendances générales entre la dégradation et de deux types de chocs aléatoires.(2) Pour les systèmes multi-composants (avec un nombre limité de composants):(a) un cadre de modélisation de processus de Markov déterministes par morceaux est développé pour traiter la dépendance de dégradation dans un système dont les processus de dégradation sont modélisées par des modèles basés sur la physique et des modèles multi-états; (b) l'incertitude épistémique à cause de la connaissance incomplète ou imprécise est considéré et une méthode volumes finis est prolongée pour évaluer la fiabilité (floue) du système; (c) les mesures d'importance de l'écart moyen absolu sont étendues pour les composants avec multiples processus concurrents dépendants de dégradation et soumis à l'entretien; (d) la politique optimale de maintenance compte tenu de l'incertitude épistémique et la dépendance de dégradation est dérivé en combinant schéma volumes finis, évolution différentielle et non-dominée de tri évolution différentielle; (e) le cadre de la modélisation de (a) est étendu en incluant les impacts des chocs aléatoires sur les processus dépendants de dégradation.(3) Pour les systèmes multi-composants (avec un grand nombre de composants), une méthode d'évaluation de la fiabilité est proposé considérant la dépendance dégradation en combinant des diagrammes de décision binaires et simulation de Monte Carlo pour réduire le coût de calcul
Analyse multivariée pour le diagnostic de l'arythmie cardiaque by Youssef Trardi( )

1 edition published in 2020 in French and held by 1 WorldCat member library worldwide

Les MCV sont parties des préoccupations sanitaires les plus pressantes et présentent la première cause de décès dans le monde. Selon l'OMS, les MCV sont à l'origine de 17,9 millions de décès dans le monde chaque année, soit 31% de l'ensemble des décès. En France, les MCV sont la deuxième cause de décès après le cancer, avec environ 150 000 décès par an. L'infarctus du myocarde, également appelé crise cardiaque, est la forme de MCV la plus meurtrière au monde. Il provoque 18 000 décès par an en France, soit 10% de la mortalité totale. Dans cette thèse, nous nous intéressons aux MCV, et plus précisément à l'une de ses principales causes, à savoir les arythmies cardiaques. Les recherches académiques et les industriels s'appuient sur les avancées technologiques pour développer des outils informatiques pour la détection de l'arythmie. Dans ce travail, nous discutons cette problématique en proposant une nouvelle stratégie de diagnostic qui permet de distinguer les sujets sains en présence de battements ectopiques des sujets atteints de la fibrillation auriculaire. Cette stratégie est basée sur l'analyse de dérivés complémentaires extraits de la série chronologique des intervalles R-R. Pour construire le modèle de diagnostic, nous avons appliqué différents algorithmes de classification, notamment les séparateurs à vaste marge et l'apprentissage multinoyaux. En outre, nous avons développé un algorithme de sélection de variables très performant, basé sur la programmation multinoyaux. L'approche développée a été validée sur différentes bases de données d'arythmies cardiaques. Les résultats obtenus démontrent l'efficacité et la robustesse de la méthode développée
contribution à la définition d'une méthodologie couplant le traitement automatique du langage naturel et l'apprentissage automatique pour réagir aux perturbations de production by Juan Pablo Usuga cadavid( )

1 edition published in 2021 in French and held by 1 WorldCat member library worldwide

In the age of Industry 4.0 (I4.0), exploiting data stored in information systems offers an opportunity to improve production systems. Datasets stored in these systems may contain patterns that machine learning (ML) models can recognise to react more effectively to future production disturbances. In the case of industrial maintenance, data are frequently collected through reports provided by operators. However, such reports are often provided using free-form text fields, resulting in complex unstructured data; therefore, they may contain irregularities such as acronyms, jargon, and typos. Furthermore, maintenance data often present asymmetrical distributions, where certain events occur more frequently than others. This phenomenon is known as class imbalance, and it can hinder the training of ML models as they tend to recognise the more frequent events better, ignoring rarer incidents. Finally, when implementing I4.0 technologies, the inclusion of humans in the decision-making process must be ensured. Otherwise, companies may be reluctant to adopt new technologies.The work presented in this thesis aims to tackle the general objective of harnessing maintenance data to react more effectively to production disturbances. To achieve this, we employed two strategies. First, we performed a systematic literature review to identify the research trends and perspectives regarding the use of ML in production planning and control. This literature analysis allowed us to understand that predictive maintenance may benefit from the unstructured data provided by operators. Additionally, their usage can contribute to the inclusion of humans in the implementation of new technologies. Second, we addressed some of the identified research gaps through case studies that employed data from real production systems. These studies harnessed the free-form text data provided by operators and presented class imbalance. Hence, the proposed case studies explored techniques to mitigate the effect of imbalanced data; moreover, we also suggested the use of a recent architecture for natural language processing called transformer
Régression polynomiale par morceaux pour la propagation de fissures by Florine Greciet( )

1 edition published in 2020 in French and held by 1 WorldCat member library worldwide

An aircraft engine is made up of several families of materials that undergo multiple degradation mechanisms from their manufacture but also during their flight cycle (take-off, landing, pressurization, pilot manoeuvring,...) or during its rest on the ground. One of these mechanisms is related to the phenomenon of fatigue which represents the degradation of a piece due to repeated cyclic stresses. This degradation results in the initiation of a crack and its propagation until the piece ruptures. The prediction of the propagation lifetime of parts is therefore a very sensitive point since it impacts both the dimensioning (step prior to the design of the pieces) and the maintenance procedures (repair or change of the piece). Propagation lifetime calculations are partly based on the laws of phenomenological evolution describing the rate of progress of the crack in a material as a function of the stress applied. In order to study these data, which are likely to be modelled continuously and which allow several propagation regimes to be observed, we propose a polynomial regression model with several regimes, subject to regularity assumptions (continuity and/or differentiability). Following this, we developed inference methods to estimate the number of regimes, transition times and parameters of each regime. These results will only be usable by the engineering office if they are obtained within reasonable calculation times, i.e. in the order of a few minutes. Each new method has therefore been designed to reduce the computation time required to estimate the model parameters. Moreover, since the number of regimes present in the data is not known a priori, the last two methods we propose do not use any a priori on this number to estimate the model parameters. The work presented in this thesis is the subject of a collaboration between the Probability and Statistics team of the Institut Elie Cartan of the University of Lorraine, the BIGS team of Inria Nancy Grand Est and Safran Aircraft Engines
Effets conjoints du vieillissement, de la maintenance et de l'hétérogénéité sur des systèmes réparables : modélisation, inférence et prise de décision by Léa Breniere( )

1 edition published in 2021 in French and held by 1 WorldCat member library worldwide

This thesis deals with reliability issues in connection with the modelling of the failure and maintenance process of repairable systems, using recurrent events. Generic virtual age models are extended by adding time-dependent covariates. These covariates take into account the observed heterogeneity between systems that are otherwise identical and independent. They can bring crucial information on the systems degradation level. We allow to use these models through two data simulation methods, as well as a parametric estimation procedure. This inferential method is numerically assessed in a thorough quality of estimation study. Then, the models are used to optimise the preventive maintenance dates along the systems' life according to the information brought by the covariates and the past repairs, allowing to reduce the maintenance costs. We go further by offering to optimise the inspections dates of the dynamic covariates. Finally, we study the implementation of multi-systems goodness-of-fit tests
moreShow More Titles
fewerShow Fewer Titles
Audience Level
Audience Level
  General Special  
Audience level: 0.00 (from 0.00 for System Rel ... to 0.00 for System Rel ...)

French (17)

English (5)