Le phénomène d'encrassement dans les échangeurs thermiques est actuellement un sujet important. En effet l'encrassement est un phénomène couteux qui provoque (directement ou indirectement via un surdimensionnement des installations) une augmentation des pertes énergétiques, une hausse de la consommation d'eau. Par effet lié, l’encrassement a un impact environnemental non négligeable dû principalement à une augmentation de l’émission de dioxyde de carbone. La détection de l'encrassement peut se faire de manière très ponctuelle en utilisant des capteurs spécifiques et coûteux ou, globalement, soit en mesurant la variation de masse de l’échangeur, soit en évaluant l’efficacité de l’échangeur à travers le coefficient de transfert – ces deux dernières méthodes exigeant des conditions de fonctionnement très particulières : l’arrêt pour la première et un fonctionnement en régime permanent pour la seconde. Les travaux présentés dans cette thèse consistent au développement d’observateurs non linéaires qui permettent de détecter l'encrassement suffisamment tôt pour mettre en place un système d'entretien efficace. Pour cela, un modèle de dimension finie d’un échangeur tubulaire à contre courant a été défini en début de thèse. Trois solutions basées sur le développement d'observateurs non linéaires de type Takagi-Sugeno appliqués au problème de détection d'encrassement dans les échangeurs thermiques sont proposées. La première consiste en une batterie d'observateurs qui estime les paramètres d'encrassement par une méthode d'interpolation. La deuxième propose un observateur polynomial de type Takagi-Sugeno en utilisant la théorie des sommes de carrés. Enfin, un observateur de type Takagi-Sugeno à entrées inconnues est développé. Une comparaison entre ces différentes méthodes est effectuée en conclusion cette thèse. / The phenomenon of fouling in heat exchangers is currently an important topic. Indeed, the fouling is a costly issue that increases the energy loss (directly or indirectly through an over-sizing of the equipment), and therefore increases the water consumption. As a side effect, fouling increases CO² consumption that leads to environmental consequences. Fouling can be detected either on local scale, using expensive and specific sensors or on global scale. Global estimation of fouling can be done by measuring the variation of the mass of the exchanger, or by estimating the efficiency of the exchanger through the transfer coefficient. These two methods require very restricting conditions: a powered exchanger to measure mass variation and a steady state exchanger to estimate the efficiency. The work introduced in this thesis deals with the development of non-linear observers that detect fouling early enough to start an efficient cleaning process. As a beginning, a finite element model of a counter current tubular exchanger was proposed. Then three approaches, based on non-linear Takagi-Sugeno observers, were suggested to detect early fouling in heat exchangers. First approach consisted in a set of observers that estimated the parameters of fouling effect through an interpolation method. The second approach proposed a polynomial Takagi-Sugeno observer, using the theory of sums of squares. Finally, a observer of Takagi-Sugeno type with unknown inputs was developed. As a conclusion, a comparison between those different methods was done.

Les personnes vivant avec une lésion de la moelle épinière (LMÉ) subissent une réduction drastique de leurs capacités sensori-motrices en dessous du niveau de la lésion. Cette pathologie influence négativement leur contrôle postural, vu comme le résultat de contributions actives, passives et neuronales. Une approche par modèle est choisie afin de représenter ces différentes contributions et mieux comprendre le contrôle postural en position assise. Deux problèmes apparaissent. Tout d’abord les modèles existants de la position assise sont très rares et inapplicables à l’étude des personnes vivant avec une LMÉ qui, dans l’absence de contrôle musculaire au niveau du tronc, vont utiliser leurs membres supérieurs pour s’équilibrer. De plus, les équations non linéaires sous forme descripteur des modèles biomécaniques existant sont linéarisées dans la majorité des cas afin de simplifier les équations obtenues ce qui n’est plus valable dans le cas de mouvements de grande amplitude. Or des travaux récents fournissent des outils pour analyser ces modèles en utilisant le formalisme de Takagi-Sugeno issu de la logique floue. Ce travail est focalisé sur l’étude du contrôle postural des personnes vivant avec une LMÉ par des outils de l’automatique non linéaire, trois objectifs sont définis : 1. Proposer des modèles biomécaniques non linéaires de la stabilité en position assise pour les personnes vivant avec une LMÉ. 2. Utiliser des méthodes d’écriture de modèle du domaine de l’automatique non linéaire afin de créer les outils nécessaires à l’estimation des variables internes (non mesurées) au modèle. 3. Améliorer la connaissance de la stabilité chez les personnes vivant avec une LMÉ par l’estimation expérimentale d’efforts internes en fonction de paramètres personnels et des conditions expérimentales. Les modèles créés dans ce manuscrit sont, à notre connaissance, les premiers à représenter la stabilité assise en considérant le déplacement des membres supérieurs. Il s’agit également des premiers modèles biomécaniques à utiliser les avancées récentes de l’automatique non linéaire pour l’estimation d’efforts internes non mesurables. / The ability to maintain stability in an unsupported sitting position often becomes challenging after an individuals is affected by a spinal cord injury (SCI). In order to better understand sitting postural control as a combination of active, passive and neuronal contribution, a modelling approach is chosen. Two problems arise; first of all, existing models of sitting control are very scarse and inapplicable to people living with a SCI who will use their upper in a compensatory strategy to balance their trunk. Moreover, the majority of biomechanical models nonlinear equations are linearized with small angle hypothesis for an easier processing which is not acceptable when movements become large. Recent advances in nonlinear control theory provide the theoretical tools for the creation of nonlinear observers using the Takagi-Sugeno fuzzy formalism. This work aims at the study of postural control for people living with a SCI by use of nonlinear tools from control theory; three main goals are formulated: 1. Creating new nonlinear biomechanical models aiming at the study of sitting stability of people living with SCI. 2. Using tools coming from nonlinear control theory to estimate model unmeasured variable 3. Improve the current knowledge on stability for people living with a SCI by estimating internal efforts depending on personal parameters in experimental conditions. The models created in this manuscript are, to the best of our knowledge, the first to consider the upper limbs displacement for the control of sitting stability and the first biomechanical models to use the recent advances in nonlinear control theory for the estimation of unmeasured variables.

本研究觀察了1992年1月20日至2003年2月28日美元兌台幣的匯率資料，分成樣本內、樣本外兩部分進行預測，此外也收集了相同時間的日圓、英鎊、港幣兌台幣的資料做比較，用Takagi-Sugeno Fuzzy﹝朱修明，2001﹞模型和Cubist決策樹模型來預測匯率。 用Takagi-Sugeno Fuzzy模型預測匯率，具有非線性模型的準確性，也兼顧了線性模型之結果簡潔易懂的特質。在變數個數少的時候，就可以達到所要求的預測準確度，此時產生的預測規則容易瞭解，歸屬度函數也易於辨別，檢定過後可知和隨機漫步模型沒有差別。 使用Cubist決策樹模型時，若產生的規則等同於隨機漫步模型，則預測準確度和隨機漫步沒有差別。但若產生出來的規則不同於隨機漫步模型時，則匯率預測準確度明顯低於隨機漫步模型。

Ce mémoire a pour objet la modélisation et la commande d'un moteur à allumage commandé. Le premier chapitre propose alors un état de l'art sur la modélisation, l'estimation et la commande dans ces derniers. Le second chapitre s'intéresse à la commande de l'admission d'air. Deux commandes y sont développées : la première est basée sur une commande linéarisante et la seconde fait appel aux outils de la commande basée sur les modèles flous de type Takagi-Sugeno. Parmi les modèles considérés, certains sont non linéaires et présentent des incertitudes de modélisation, et d'autres sont à retards variables. Le troisième chapitre présente alors un certain nombre de développements théoriques concernant deux classes de modèles flous : ceux à retards variables sur l'état, et ceux avec incertitudes. Des conditions de stabilisation sans et avec un observateur d'état sont également données. Le quatrième chapitre traite alors de la commande et de l'observation par retour d'état basées sur des modèles flous de type Takagi-Sugeno et/ou par modes glissants, et de leur validation sur le modèle développé. Une régulation de richesse, une commande robuste du couple moteur moyen, un observateur à entrées inconnues du couple instantané, et des essais de commande globale sont présentés.

No description available.

In this thesis we have considered a cognitive radio network (CRN) with a pair of primary user (PU) and secondary user (SU) in spectrum sharing networks in path-loss and without path-loss propagation environments under identically distributed m-Nakagami fading channel. The thesis consists of three parts. In the first part we propose an optimized Takagi-Sugeno Fuzzy Inference System (FIS) based power control strategy in cognitive radio networks (CRN) in spectrum sharing network in without path-loss propagation environment. The second part proposes an optimized Takagi-Sugeno FIS based power control strategy in cognitive radio networks in spectrum sharing network in path-loss propagation environment. For without path-loss propagation environment the proposed FIS takes the interference channel gain ratio between SU transmitter (CUtx) and PU receiver (PUrx) and Signal to Noise Ratio (SNR) towards PU transmitter (PUtx) as antecedents and outputs the power scaling factor for SU. For path-loss propagation environment the proposed FIS takes the relative distance ratio between CUtx and PUrx and SNR towards PUtx as antecedents and outputs the power scaling factor for SU. The output power scaling factor is used to vary the transmit power of SU such that it does not degrade the quality of service (QoS) of PU link. The third part presents an implementation of orthogonal frequency division multiplexing (OFDM) transmission technique in CRN. The OFDM technique has intellectual attractive features like coping with the inter symbol interference (ISI), while providing increasing spectral efficiency and improved performance. This can be used in emergency conditions where transmission requires reliability and high data rate. The OFDM transmission technique is applied towards SU transmitter in CRN, which enables SU to utilize the spectrum efficiently under various fading environments. Spectrum sharing networks in with and without path-loss propagation environments and OFDM transmission were tested for bit error rate (BER) performance after fading effects from m-Nakagami fading channel. We conclude that by applying Takagi-Sugeno Fuzzy Inference System (FIS) based power control strategy we can improve the BER performance of PU when compared with no power control strategy and with other fuzzy based power control technique. OFDM transmission technique gives us better data rate and slightly improved BER in CRN hence making it suitable for use in emergency conditions. / mobile: 0735032048 (Muneeb Goraya)

Made available in DSpace on 2014-06-11T19:30:32Z (GMT). No. of bitstreams: 0 Previous issue date: 2009-08-21Bitstream added on 2014-06-13T21:01:21Z : No. of bitstreams: 1 faria_fa_dr_ilha.pdf: 872755 bytes, checksum: d14b68025a60624f43ac43ae6b5e757e (MD5) / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) / Já é conhecido da teoria de controle clássico, que a realimentação derivativa pode ser útil e em alguns casos, essencial para a obtenção do desempenho desejado. Além disso, existem problemas práticos em que a derivada dos estados é mais fácil de se obter do que os sinais dos estados. Por exemplo, nos seguintes problemas: controle de vibrações de sistemas mecânicos, sistemas de suspensão de automóveis, controle de vibrações em pontes suspensas e controle de vibrações na aterrissagem de aviões. O principal sensor usado nesses problemas é o acelerômetro. A partir do sinal do acelerômetro é possível obter com precisão a velocidade, mas não o deslocamento. Definindo o deslocamento e a velocidade como as variáveis de estado, então apenas os sinais da derivada dos estados estão disponíveis para a realimentação. A realimentação derivativa foi usada recentemente no projeto de sistemas lineares. Os procedimentos consideram o problema de alocação de pólos (ABDELAZIZ; VALÁŠEK, 2004, 2005) e o projeto de reguladores lineares quadráticos (DUAN; NI; KO, 2005). Infelizmente esses resultados não podem ser usados em sistemas incertos ou sujeitos à falhas estruturais. Para resolver esse caso, projetos baseados em DesigualdadesMatriciais Lineares (LMIs) foram propostos em (FARIA, 2005; ASSUNÇÃO et al., 2007e; FARIA; ASSUNÇÃO; TEIXEIRA, 2009a; CARDIM et al., 2008; FARIA et al., 2009c). LMIs podem ser facilmente resolvidas com softwares de otimização, tais como os pacotes “LMI control toolbox” e “SeDuMi” do MATLAB. Este trabalho aborda o projeto de controladores para sistemas descritores usando a realimentação da derivada dos estados. São apresentadas condições necessárias e suficientes baseadas em LMIs para a estabilidade assintótica de sistemas descritores. Como a estabilidade nem sempre é suficiente, posteriormente... / From classical control theory, it is well-known that state-derivative feedback can be very useful, and even in some cases essential to achieve a desired performance. Moreover, there exist some practical problems where the state-derivative signals are easier to obtain than the state signals. For instance, in the following applications: suppression of vibration in mechanical systems, control of car wheel suspension systems, vibration control of bridge cables and vibration control of landing gear components. The main sensor used in these problems is the accelerometer. In this case, from the signals of the accelerometers it is possible to reconstruct the velocities with a good precision but not the displacements. Defining the velocities and displacement as the state variables, then one has available for feedback the state-derivative signals. Recent researches about state-derivative feedback design for linear systems have been presented. The procedures consider, for instance, the pole placement problem (ABDELAZIZ; VALÁŠEK, 2004, 2005), and the design of a Linear Quadratic Regulator (DUAN; NI; KO, 2005). Unfortunately these results are not applied to the control of uncertain systems or systems subject to structural failures. For this case, Linear Matrix Inequalities (LMIs) based designs were recently proposed in (FARIA, 2005; ASSUNÇÃO et al., 2007e; FARIA; ASSUNÇÃO; TEIXEIRA, 2009a; CARDIM et al., 2008; FARIA et al., 2009c). LMIs can be easily solved with optimization softwares, such as the packages “LMI control toolbox” and “SeDuMi” of MATLAB. This work focuses control designs for descriptor systems using the state-derivative feedback. Necessary and sufficient LMI-based conditions for asymptotically stability of descriptor systems are presented. As the stability of a control system is insufficient to obtain a suitable performance... (Complete abstract click electronic access below)

Made available in DSpace on 2014-06-11T19:22:31Z (GMT). No. of bitstreams: 0 Previous issue date: 2009-02-20Bitstream added on 2014-06-13T18:49:28Z : No. of bitstreams: 1 silva_erp_me_ilha.pdf: 509609 bytes, checksum: 37c083ec485dbc3db2cf95730b404b40 (MD5) / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) / Uma técnica de projeto de controladores robustos para sistemas não-lineares contínuos no tempo é proposta neste trabalho. É suposto que a planta não-linear está sujeita a falhas estruturais, que podem ser consideradas como incertezas politópicas. Os sistemas não-lineares são representados por modelos fuzzy propostos por Takagi-Sugeno e uma formulação para o tratamento das incertezas politópicas é apresentado para o projeto dos controladores. Este trabalho aborda projetos de controle usando a realimentação dos estados e a realimentação da derivada dos estados. O projeto do controlador é realizado através de condições baseadas em Desigualdades Matriciais Lineares (em inglês, Linear Matrix Inequalities (LMI)), que podem ser resolvidas facilmente utilizando técnicas de programação convexa. Essa metodologia permite a inclusão de restrições de desempenho no projeto, tais como: taxa de decaimento e restrição na entrada. Ao final, exemplos numéricos e suas simulações ilustram a eficiência da técnica proposta. / A technique of robust controllers design for nonlinear continuous-time systems is proposed in this work. It is supposed that the nonlinear plant is subject to structural failures, which can be considered as polytope uncertainties. The nonlinear systems are represented through fuzzy models proposed by Takagi-Sugeno and a formulation for the treatment of polytope uncertain is presented for the controllers design. This work focuses control designs using state feedback and state-derivative feedback. The controllers design is made through conditions based in Linear Matrix Inequalities (LMIs), which can be easily solved using convex programming techniques. This methodology allows the inclusion of performance restrictions on design, such as: decay rate and input constraint. Finally, numeric examples and their simulations show the efficiency of the proposed method.

Orientador: Marcelo Carvalho Minhoto Teixeira / Coorientador: Rodrigo Cardim / Banca: Edvaldo Assunção / Banca: Alfredo Del Sole Lordelo / Resumo: Neste trabalho é realizada a análise da estabilidade de sistemas fuzzy Takagi-Sugeno (TS) contínuos no tempo, através de Funções de Lyapunov Fuzzy (FLF), Funções de Lyapunov Metzler (FLM) e de Funções de Lyapunov Fuzzy-Metzler (FLFM) introduzida nesta disser- tação. Novas propostas são feitas a partir destas análises, sendo apresentadas condições su- ficientes para a estabilidade assintótica destes sistemas no sentido de Lyapunov. As soluções obtidas são baseadas em desigualdades lineares matriciais (LMIs, do inglês Linear Matrix Ine- qualities) e dependem da solução de um conjunto de desigualdades de Lyapunov-Metzler, que podem ser de difícil solução. Então, foram apresentadas condições de estabilidade baseadas em uma subclasse de matrizes de Metzler que, quando factíveis, podem ser resolvidas através de LMIs com a necessidade de uma busca unidimensional. Foram propostos métodos que genera- lizam os já existentes na literatura, baseados em FLF, para a estabilidade assintótica dos sistemas fuzzy TS / Abstract: This work addresses the stability analysis of Takagi-Sugeno (TS) fuzzy systems via Fuzzy Lyapunov Functions (FLF), Metzler Lyapunov Functions (MLF) and Fuzzy-Metzler Lyapunov Functions (FMLF) that was proposed in this dissertation. New proposals are made from these analyses, and sufficient conditions for asymptotic stability of these systems in the sense of Lyapunov are presented. The results obtained are based on LMIs (Linear Matrix Inequalities) and depend on the solutions of a set of Lyapunov-Metzler inequalities, that are usually difficult to solve. Then, conditions for stability based on a subclass of Metzler matrices that, when feasible, can be described by a set of LMIs with an unidimensional search, are presented. The proposed methods generalize the similar methods available in the literature, based on FLF, for the asymptotic stability of TS fuzzy systems / Mestre

Diese Arbeit leistet einen Beitrag zur Modellierung und Regelung nichtlinearer dynamischer Systeme mit Hilfe von Takagi-Sugeno Fuzzy-Systemen. Zielsetzung ist es, zur Modellierung von nichtlinearen dynamischen Systemen fuzzy-verknüpfte, lokal gültige lineare Modelle einzusetzen, deren Systemverhalten lokal dem der nichtlinearen Systeme entspricht. Ein Schwerpunkt der Arbeit liegt auf der Ableitung einer geeigneten Partition des Eingangsgrößenbereichs, d.h. auf der Ableitung optimaler Klassen, bestehend aus Gültigkeitsbereichen und lokalen Modellen, die sich durch geringe Redundanz auszeichnen. Dazu wird ein adaptives Verfahren hergeleitet, das die Anpassung der Gültigkeitsbereiche durch die Bewertung der Ähnlichkeit zwischen den lokalen Modellen und dem zu approximierenden Referenzsystem durchführt. Die Leistungsfähigkeit des Verfahrens wird anhand der Simulation und der Regelung des Antriebsstrangs einer Windkraftanlage mit Überlagerungsgetriebe demonstriert. Hierzu wird für die nichtlineare Abhängigkeit des Rotormoments eine Modellstruktur abgeleitet, die zur Modellierung mit lokalen linearen Modellen verwendet werden kann. Zur Regelung wird auf das Prinzip der LQI-Zustandsregler zurückgegriffen, wobei der Einfluss der Windgeschwindigkeit mit dem Prinzip der Störgrößenaufschaltung reguliert wird. Um dies realisieren zu können, werden die Werte der nicht exakt messbaren Windgeschwindigkeit durch einen Störgrößenbeobachter bereitgestellt, der ebenfalls auf dem Prinzip der lokalen linearen Modelle basiert.