Community detection in networks is an important area of current research with many applications. Finding community structures is a challenging task and despite significant effort no satisfactory method has been found. Different methods find different communities in the same network and with different computational requirements. To counter this problem, several different methods are often used and the results compared manually. In this thesis, we present three different methods to instead merge the results from different methods (or several runs from the same algorithm) to find better estimates of the community structure. Another problem in practical applications is noisy and imperfect networks with missing and false edges. These imperfections are natural results from the methods used to map the network structure and are often difficult to eliminate. In this thesis, we apply a Monte Carlo-sampling method in combination with the introduced methods for merging community detection results to find community structures in such networks. The method is tested by simulation studies on both real-world networks and synthetic networks with generated uncertainties and imperfections. We finally demonstrate how it is possible to generate confidence levels of the obtained community structure from the merging methods. This allows for a qualitative comparison of the robustness and significance of the network clustering. / Identifikation av grupperingar i nätverk är ett viktigt område inom aktuell forskning med många olika tillämpningsområden. Att finna grupperingar är ofta svårt och trots betydande ansträngningar har ingen tillfredsställande metod hittats. Olika metoder finner ofta olika grupperingar i samma nätverk och kräver varierande beräkningskraft. För att hantera dessa problem används ofta flera metoder vartefter resultaten jämförs manuellt. I detta examensarbete presenterar vi tre olika metoder att istället slå samman resultat från olika metoder (eller fler körningar från samma algoritm) för att hitta bättre uppskattningar av grupperingarna. Ett annat problem i praktiska tillämpningar är brus och ofullständiga nätverk med saknade och falska kanter. Dessa brister är naturliga resultat från de metoder som används för att kartlägga nätverketstrukturen och det är ofta svåra att eliminera dessa. I detta examensarbete använder vi Monte Carlo-metoder i kombination med de introducerade metoderna för att slå samman funna grupperingar för att hitta grupperingar i det osäkra nätverket. Vi testar metoden genom simuleringstudier på både verkliga och syntetiska nätverk med genererade osäkerheter och brister. Slutligen demostrerar vi hur det är möjligt att skapa konfidensnivåer för noder i grupperingar med hjälp av metoderna för sammanslagning. Detta möjliggör en kvalitativ jämförelse av stabilitet och signifikans av identifierade nätverksgrupperingar.

This dissertation seeks to merge the ideas from robust control theory such as H-Infinity control design and the Small Gain Theorem, L stability theory and Lyapunov stability from nonlinear control, and recent theoretical achievements in adaptive control. The fusion of frequency domain and linear time domain ideas allows the derivation of an H-Infinity Norm Minimization Approach (H-Infinity-NMA) for adaptive control architecture that permits a control designer to simplify the adaptive tuning process and tune the uncertainty compensation characteristics via linear control design techniques, band limit the adaptive control signal, efficiently handle redundant actuators, and handle unmatched uncertainty and matched uncertainty in a single design framework. The two stage design framework is similar to that used in robust control, but without sacrificing performance. The first stage of the design considers an ideal system with the system uncertainty completely known. For this system, a control law is designed using linear H-Infinity theory. Then in the second stage, an adaptive process is implemented that emulates the behavior of the ideal system. If the linear H-Infinity design is applied to control the emulated system, it then guarantees closed loop system stability of the actual system. All of this is accomplished while providing notions of transient performance bounds between the ideal system and the true system. Extensions to the theory include architectures for a class of output feedback systems, limiting the authority of an adaptive control system, and a method for improving the performance of an adaptive system with slow dynamics without any modification terms. Applications focus on using aerodynamic flow control for aircraft flight control and the Crew Launch Vehicle.

No description available.

Several techniques were studied to select and prioritize technologies for a complex system. Based on the findings, a method called Pareto Optimization and Selection of Technologies (POST) was formulated to efficiently explore the combinatorial technology space. A knapsack problem was selected as a benchmark problem to test-run various algorithms and techniques of POST. A Monte Carlo simulation using the surrogate models was used for uncertainty quantification. The concepts of graph theory were used to model and analyze compatibility constraints among technologies. A probabilistic Pareto optimization, based on the concepts of Strength Pareto Evolutionary Algorithm II (SPEA2), was formulated for Pareto optimization in an uncertain objective space. As a result, multiple Pareto hyper-surfaces were obtained in a multi-dimensional objective space; each hyper-surface representing a specific probability level. These Pareto layers enabled the probabilistic comparison of various non-dominated technology combinations. POST was implemented on a technology exploration problem for a 300 passenger commercial aircraft. The problem had 29 identified technologies with uncertainties in their impacts on the system. The distributions for these uncertainties were defined using beta distributions. Surrogate system models in the form of Response Surface Equations (RSE) were used to map the technology impacts on the system responses. Computational complexity of technology graph was evaluated and it was decided to use evolutionary algorithm for probabilistic Pareto optimization. The dimensionality of the objective space was reduced using a dominance structure preserving approach. Probabilistic Pareto optimization was implemented with reduced number of objectives. Most of the technologies were found to be active on the Pareto layers. These layers were exported to a dynamic visualization environment enabled by a statistical analysis and visualization software called JMP. The technology combinations on these Pareto layers are explored using various visualization tools and one combination is selected. The main outcome of this research is a method based on consistent analytical foundation to create a dynamic tradeoff environment in which decision makers can interactively explore and select technology combinations.

Le processus d'évaluation des risques technologiques, notamment liés au Transport de Matières Dangereuses (TMD), consiste, quand un événement accidentel se produit, à évaluer le niveau de risque potentiel des zones impactées afin de pouvoir dimensionner et prendre rapidement des mesures de prévention et de protection (confinement, évacuation...) dans le but de réduire et maitriser les effets sur les personnes et l'environnement. La première problématique de ce travail consiste donc à évaluer le niveau de risque des zones soumises au transport des matières dangereuses. Pour ce faire, un certain nombre d'informations sont utilisées, comme la quantification de l'intensité des phénomènes qui se produisent à l'aide de modèles d'effets (analytique ou code informatique). Pour ce qui concerne le problème de dispersion de produits toxiques, ces modèles contiennent principalement des variables d'entrée liées à la base de données d'exposition, de données météorologiques,… La deuxième problématique réside dans les incertitudes affectant certaines entrées de ces modèles. Pour correctement réaliser une cartographie en déterminant la zone de de danger où le niveau de risque est jugé trop élevé, il est nécessaire d'identifier et de prendre en compte les incertitudes sur les entrées afin de les propager dans le modèle d'effets et ainsi d'avoir une évaluation fiable du niveau de risque. Une première phase de ce travail a consisté à évaluer et propager l'incertitude sur la concentration qui est induite par les grandeurs d'entrée incertaines lors de son évaluation par les modèles de dispersion. Deux approches sont utilisées pour modéliser et propager les incertitudes : l'approche ensembliste pour les modèles analytiques et l'approche probabiliste (Monte-Carlo) qui est plus classique et utilisable que le modèle de dispersion soit analytique ou défini par du code informatique. L'objectif consiste à comparer les deux approches pour connaitre leurs avantages et inconvénients en termes de précision et temps de calcul afin de résoudre le problème proposé. Pour réaliser les cartographies, deux modèles de dispersion (Gaussien et SLAB) sont utilisés pour évaluer l'intensité des risques dans la zone contaminée. La réalisation des cartographies a été abordée avec une méthode probabiliste (Monte Carlo) qui consiste à inverser le modèle d'effets et avec une méthode ensembliste générique qui consiste à formuler ce problème sous la forme d'un ensemble de contraintes à satisfaire (CSP) et le résoudre ensuite par inversion ensembliste. La deuxième phase a eu pour but d'établir une méthodologie générale pour réaliser les cartographies et améliorer les performances en termes de temps du calcul et de précision. Cette méthodologie s'appuie sur 3 étapes : l'analyse préalable des modèles d'effets utilisés, la proposition d'une nouvelle approche pour la propagation des incertitudes mixant les approches probabiliste et ensembliste en tirant notamment partie des avantages des deux approches précitées, et utilisable pour n'importe quel type de modèle d'effets spatialisé et statique, puis finalement la réalisation des cartographies en inversant les modèles d'effets. L'analyse de sensibilité présente dans la première étape s'adresse classiquement à des modèles probabilistes. Nous discutons de la validité d'utiliser des indices de type Sobol dans le cas de modèles intervalles et nous proposerons un nouvel indice de sensibilité purement intervalle cette fois-ci. / When an accidental event is occurring, the process of technological risk assessment, in particular the one related to Dangerous Goods Transportation (DGT), allows assessing the level of potential risk of impacted areas in order to provide and quickly take prevention and protection actions (containment, evacuation ...). The objective is to reduce and control its effects on people and environment. The first issue of this work is to evaluate the risk level for areas subjected to dangerous goods transportation. The quantification of the intensity of the occurring events needed to do this evaluation is based on effect models (analytical or computer code). Regarding the problem of dispersion of toxic products, these models mainly contain inputs linked to different databases, like the exposure data and meteorological data. The second problematic is related to the uncertainties affecting some model inputs. To determine the geographical danger zone where the estimated risk level is not acceptable, it is necessary to identify and take in consideration the uncertainties on the inputs in aim to propagate them in the effect model and thus to have a reliable evaluation of the risk level. The first phase of this work is to evaluate and propagate the uncertainty on the gas concentration induced by uncertain model inputs during its evaluation by dispersion models. Two approaches are used to model and propagate the uncertainties. The first one is the set-membership approach based on interval calculus for analytical models. The second one is the probabilistic approach (Monte Carlo), which is more classical and used more frequently when the dispersion model is described by an analytic expression or is is defined by a computer code. The objective is to compare the two approaches to define their advantages and disadvantages in terms of precision and computation time to solve the proposed problem. To determine the danger zones, two dispersion models (Gaussian and SLAB) are used to evaluate the risk intensity in the contaminated area. The risk mapping is achieved by using two methods: a probabilistic method (Monte Carlo) which consists in solving an inverse problem on the effect model and a set-membership generic method that defines the problem as a constraint satisfaction problem (CSP) and to resolve it with an set-membership inversion method. The second phase consists in establishing a general methodology to realize the risk mapping and to improve performance in terms of computation time and precision. This methodology is based on three steps: - Firstly the analysis of the used effect model. - Secondly the proposal of a new method for the uncertainty propagationbased on a mix between the probabilistic and set-membership approaches that takes advantage of both approaches and that is suited to any type of spatial and static effect model. -Finally the realization of risk mapping by inversing the effect models. The sensitivity analysis present in the first step is typically addressed to probabilistic models. The validity of using Sobol indices for interval models is discussed and a new interval sensitivity indiceis proposed.

Des projections réalisées sur les perspectives démographiques et financières de la dépendance prévoient, en France, une nette augmentation de la population des plus de 80 ans, accompagnée d'une multiplication par 2 du nombre de personnes âgées dépendantes entre 2010 et 2060. Afin de gérer l'augmentation du nombre de personnes âgées dépendantes, les EHPAD (Etablissement d'Hébergement pour Personne Agées Dépendantes) sont appelés à améliorer la prise en charge des résidents et à améliorer les conditions de travail du personnel soignant. C'est dans ce contexte, que nous avons développé un système de supervision permettant de détecter, via un ensemble de capteurs, des évolutions du comportement ou encore, le comportement anormal d'une personne âgée. La détection des comportements anormaux dans le cadre de la supervision est un sujet de recherche qui a été largement étudié dans la littérature ; ce qui n'est tout de même pas le cas de l'analyse des variations des activités de la vie de tous les jours, prenant en compte les spécificités du comportement de la personne au cours du temps. Ainsi, nous avons proposé un modèle de Markov, permettant d'apprendre, avec le moins d'a priori possible, le modèle de comportement habituel au sein de la chambre. Le modèle proposé a été testé sur des données acquises en Living Lab (GIS-Madonah). Nous avons également proposé une nouvelle approche pour calculer la distance entre deux modèles de Markov, afin d'évaluer l'évolution du comportement au cours du temps. Ces méthodes devront permettre, non seulement de déterminer la probabilité du comportement actuel de la personne par rapport à son comportement habituel ; mais également, de détecter des évolutions lentes du comportement de la personne. / Due to demographic changes, it is expected that the number of French having over 80 years will increase drastically and the number of dependent elderly people will grow twice between 2010 and 2060. To manage this increasing number of dependent elderly person, nursing homes are required to improve the care of residents and to improve the working conditions of health workers. In this context, we plan to develop a monitoring system, based on a set of sensors, to detect modifications in the behavior of a person, and unusual behavior. Detection of abnormal activities in smart homes is an important topic of research, unlike the detection of the evolutions of behavior, which take into account the specifics activities of the person in time. Thus, we proposed a Markov model which allow to learn the usual behavior in the room, with a reduced number of a priori. The model is try on data acquired on a Living Lab (GIS Madonah). We proposed a new method to compute the distance between two Markov models, to estimate the evolution of the behavior. These methods allow to compute the probability of the current activities with the usual behavior, and the slow evolutions of the behavior.

Orientador: Marcelo Carvalho Minhoto Teixeira / Resumo: Neste trabalho são propostos novos métodos de controle chaveado para uma classe de sistemas não lineares incertos utilizando a decomposição em soma de quadrados. Inicialmente é apresentada uma revisão dos conceitos e projetos de controladores baseados em desigualdades matriciais lineares (do inglês Linear Matrix Inequalities - LMIs) e a decomposição em soma de quadrados (do inglês Sum of Squares - SOS), buscando evidenciar as diferenças e vantagens das metodologias para a área de controle. Comumente são utilizados modelos fuzzy para realizar a análise da estabilidade e projeto de controladores para sistemas não lineares, e estes modelos podem ser classificados de acordo com a parte consequente linear ou polinomial. Busca-se neste trabalho evidenciar as diferenças entre os dois modelos fuzzy e a metodologia para projeto de controladores. Para o caso de sistemas cujas dinâmicas podem ser descritas apenas por funções polinomiais, serão consideradas incertezas politópicas. Então, visando flexibilizar o projeto utilizando um controlador composto por um único ganho polinomial e aumentar a região de factibilidade, são propostos controladores com ganhos polinomiais chaveados. O objetivo desta lei de chaveamento é minimizar a derivada da função de Lyapunov empregada no projeto. Considerando uma classe de sistemas não lineares mais geral, são propostos controladores com ganhos chaveados para modelos fuzzy polinomiais. A metodologia proposta não necessita do conhecimento das funções de ... (Resumo completo, clicar acesso eletrônico abaixo) / Abstract: In this manuscript new control methods are proposed for a class of uncertain nonlinear systems using a sum of squares decomposition. Initially is presented a revision of concepts and control design procedures based on Linear Matrix Inequalities (LMIs) and on sum of squares (SOS) evidencing the differences and advantages of these methodologies in the control system design. Fuzzy models are commonly used to perform stability analysis and controller design for nonlinear systems, and can be classified by a linear or polynomial consequent model. A goal of this dissertation is to compare these two methodologies in the control system design of a class of uncertain nonlinear systems. For the case of systems whose dynamics can be described only by polynomial functions will be also considered polytopic uncertainty. Therefore, in order to make the design more flexible than that obtained with only one controller with polynomial gain and increase the feasibility region, a new procedure for designing controllers with switched polynomial gains is proposed. The purpose of this switching law is to minimize the time derivative of the Lyapunov function employed in the design. For a more general class of nonlinear systems, controllers with switched gains for polynomial fuzzy models are proposed. The proposed methodology does not require the knowledge of the membership functions for an implementation of the control law. This fact is an important advantage over the many methods that consider avail... (Complete abstract click electronic access below) / Mestre

O estudo da estabilidade e desempenho em sistemas de controle é um tópico relevante na teoria de sistemas. Quando são assumidas incertezas no modelo da planta, existe uma maior dificuldade para garantir um nível de desempenho adequado do sistema dinâmico e a estabilidade pode ser comprometida. Neste trabalho são utilizados um regulador linear quadrático robusto e um filtro de Kalman robusto combinados em uma única formulação para tratar de sistemas dinâmicos incertos em tempo real. O caso de estudo selecionado é o pêndulo invertido. Seus principais desafios de controle encontrados na literatura: estabilização, seguimento e levantamento-captura, serão considerados. Os algoritmos utilizados são motivados pelo fato de que problemas estocásticos podem ser resolvidos por meio de argumentos determinísticos, baseados nos conceitos de função penalidade e mínimos quadrados regularizados. Desta forma, é possível a obtenção do melhor desempenho em contrapartida à máxima influência de incerteza admissível. A análise de desempenho do controlador robusto é realizada por meio de ensaios práticos incluindo incertezas na planta, ruído nos sensores e distúrbios no sinal de controle do pêndulo. / The study of stability and performance in control systems is a relevant topic in systems theory. When uncertainties are considered in the model of the plant, there is a greater difficulty in ensuring an appropriate performance level of the dynamic system, plus, the stability could be compromised as well. In this dissertation a robust linear quadratic regulator and a robust Kalman filter are used in a unified manner to deal with uncertain dynamic systems in real time. The selected case study is the inverted pendulum. Its main control challenges found in the literature will be considered: stabilization, tracking and catching swing-up. The used algorithms are motivated by the fact that stochastic problems can be solved through deterministic arguments based on the concepts of penalty function and regularized least-squares. Thus, it is possible to obtain an optimal performance for the maximum acceptable uncertainty. The performance analysis of the robust control is carried out by practical experiments including uncertainties in the plant, noise in the sensors and disturbance in the pendulum control signal.

Orientador: Paulo Augusto Valente Ferreira / Tese (doutorado) - Universidade Estadual de Campinas, Faculdade de Engenharia Eletrica e de Computação / Made available in DSpace on 2018-08-06T08:04:01Z (GMT). No. of bitstreams: 1 Prado_MarciaLissandraMachado_D.pdf: 1397184 bytes, checksum: 5c23769b43bb76b58c39a5fb95adaaee (MD5) Previous issue date: 2006 / Resumo: Uma abordagem baseada em análise intervalar para o projeto de controladores por realimentação de estados robusta é proposta. Demonstra-se que quando especificações para alocação de pólos são representadas por conjuntos espectrais de polinômios intervalares, o problema do projeto por realimentação de estados robusta pode ser completamente formulado e resolvido no contexto de conceitos e métodos de análise intervalar. Representações poliédricas convexas de uma classe de controladores por realimentação de estados robusta satisfazendo a uma equação de Ackerman intervalar são derivadas. Um procedimento de projeto baseado em programação não-linear que objetiva a maximização da não-fragilidade do controlador robusto resultante é introduzido. Para sistemas multivariáveis é proposta uma abordagem por alocação de pólos utilizando a equação de Sylvester intervalar e técnicas de resolução baseadas em intervalos modais. Exemplos numéricos ilustram o projeto de controladores por realimentação de estados obtidos a partir da abordagem por análise intervalar proposta / Abstract: An interval analysis approach for the design of robust state feedback controllers is proposed. It is shown that when regional pole placement specifications are represented as spectral sets of interval polynomials, the robust state feedback design problem can be entirely formulated and solved in the context of the concepts and methods of interval analysis. Explicit convex polyhedral representations of a class of robust state feedback controllers satisfying an interval Ackerman¿s equation are derived. A design procedure based on nonlinear programming which aims at maximizing the non-fragility of the resulting robust controller is introduced. In the case multivariable systems is proposed an approach based on pole placement which employs an interval Sylvester equation and modal intervals techniques. Numerical examples illustrate the design of robust state feedback controllers through the interval analysis approaches proposed. / Doutorado / Automação / Doutor em Engenharia Elétrica

This thesis presents the implementation of a knowledge base system for rough sets [Paw92]within the logic programming framework. The combination of rough set theory with logic programming is a novel approach. The presented implementation serves as a prototype system for the ideas presented in [VDM03a, VDM03b]. The system is available at "http://www.ida.liu.se/rkbs". The presented language for describing knowledge in the rough knowledge base caters for implicit definition of rough sets by combining different regions (e.g. upper approximation, lower approximation, boundary) of other defined rough sets. The rough knowledge base system also provides methods for querying the knowledge base and methods for computing quantitative measures. We test the implemented system on a medium sized application example to illustrate the usefulness of the system and the incorporated language. We also provide performance measurements of the system.