L'artichaut et le cardon, appartenant à la famille des Asteraceae (Compositae), sont des plantes pérennes herbacées natives du bassin méditerranéen, et qui sont traditionnellement cultivées comme plantes maraîchères, respectivement pour leurs têtes et leurs cardes. L'Italie est le pays possédant la plus importante collection de germoplasmes autochtones d'artichaut. Dans le centre de l'Italie, le type Romanesco est étendu. Ces dernières années, le développement des techniques in vitro a permis la multiplication de l'artichaut Romanesco et sa rapide expansion. Le clone Romanesco C3 a ainsi remplacé de nombreuses races locales de Romanesco, contribuant de la sorte à une érosion significative des ressources génétiques locales. Concernant le germoplasme de cardon cultivé, seules quelques études sont disponibles sur son identification et sa caractérisation génétique. Le cardon sauvage, n'est pas du tout cultivé; il est davantage considéré comme une adventice dans le paysage italien. La grande variabilité existant entre les espèces de Cynara n'est pas correctement décrite et il existe plusieurs cas d'homonymes. Le secteur de l'artichaut italien a du faire face ces dernières années à une crise causée principalement par l'apparition sur le marché de produits étrangers et par les coûts élevés de mise en culture et de récolte. Afin de surmonter cette crise, de nombreuses valorisations du Cynara ont été envisagées. Ces potentielles applications pour la culture ont pu voir le jour grâce au support de l'Union européenne pour la recherche sur les co-poduits issus de l'agriculture, et ont mené à un intérêt croissant pour la biomasse entière d'artichaut. Considérant ces remarques préliminaires, une stratégie, qui permettrait de valoriser le germoplasme italien par la production concomitante de biomasse et de biocomposés, a été mise en place durant ces trois ans de doctorat. Le premier objectif de ce travail de doctorat consistait en i) la caractérisation agro-morphologique du germoplasme italien par le biais de descripteurs UPOV, ii) l'évaluation de la variabilité génétique à l'intérieur et entre les races/clones et iii) l'identification et la préservation des ressources génétiques pour le développement de futurs programmes d'amélioration des plantes. Suite à cette caractérisation, trois génotypes ont été sélectionnés et enregistrés sous les noms de Michelangelo, Donatello and Raffaello. Afin d'analyser le germoplasme italien de Cynara d'un point de vue de la biomasse, différents traits expliquant la vigueur de la plante et la production de matière sèche ont été considérés. Le rendement en biomasse aérienne s'est révélé très élevé, soulignant la possibilité d'utiliser cette culture comme matière première industrielle. Un point particulier du programme de thèse était de mettre au point les méthodes d'extraction de biocomposés et les techniques d'analyse afin d'optimiser le rendement en polyphénols à partir de la biomasse de Cynara à l'échelle laboratoire. L'ASE a été reconnue comme étant la meilleure technique. De plus, les cinétiques de production de biomasse et de biocomposés ont été évaluées et le stade physiologique optimal pour collecter le matériel végétal en champ a été identifié. La caractérisation biochimique a été réalisée grâce aux méthodes mises au point et en collectant le matériel végétal au stade physiologique optimal identifié afin de distinguer les génotypes les plus appropriés pour la production de biocomposés. Le dernier point du programme de thèse était centré sur le développement d'une technique alternative de production de biomasse et de biocomposés en conditions de croissance sous serre. Les résultats obtenus mettent en exergue la possibilité d'utiliser avec succès certains génotypes de Cynara pour la production de biomasse et de biocomposés. La perspective réelle d'utiliser certains génotypes d'artichaut pour une double valorisation alimentaire et non-alimentaire a ainsi été soulignée. / Globe artichoke and cardoon, belonging to the Asteraceae (Compositae) family, are herbaceous perennial plants native to the Mediterranean area, which are traditionally grown as vegetables for the heads and the fleshy petiole leaves, respectively. Italy is the richest reserve of globe artichoke autochthonous germplasm, which is vegetatively propagated and well adapted to the different pedoclimatic conditions of the Country. In Central Italian environments, the Romanesco type is widespread. In the last years, the development of in vitro technologies allowed the propagation of Romanesco globe artichoke type and its rapid expansion. As a result, the micropropagated Romanesco clone C3 has replaced many landraces traditionally grown in the Latium Region and has led either to a significant erosion of local genetic resources. As regards Italian cultivated cardoon germplasm, there are few studies on its genetic characterization and identification and there is a lack of information about the genetic variability existing within and among accessions. For the wild cardoon, no specialized crop is present and it represents mainly a weed in Italian environments. The great variability existing in Cynara spp. has not been described, the nomenclature of Italian germplasm is not always very clear since there are many cases of homonyms. In addition, Italian globe artichoke sector is facing a crisis due principally to the appearance on the market of foreign products and to the high labor cost required for crop cultivation and harvesting. In order to overcome this crisis several possible industrial uses were considered for the species. Considering these preliminary remarks, a strategy for valorizing Italian germplasm using biomass and biocompound production has been carried out during the three years of PhD program. The first objective of PhD work consisted in (i) characterizing agro-morphologically Italian germplasm using UPOV descriptors, (ii) assessing the genetic variability existing within and among landraces/clones and (iii) identifying and preserving genetic resources for the development of future plant breeding programs. As a result of this characterization, three genotypes have been selected and registered under the names of Michelangelo, Donatello and Raffaello. In order to analyze Italian Cynara spp. germplasm also from a biomass point of view, different traits explaining plant vigor and dry matter production have been considered. The aerial biomass yield resulted very high underlining the possibility of using this crop as raw industrial material. A focal point of PhD program was to set up biocompound extraction methods and analysis techniques to optimize polyphenol recovery from biomass of Cynara spp at a laboratory scale. In particular, ASE was found as the best extraction technique which allows to reduce extraction time and solvent consumption, increase nutraceutical yield and improvement of extract quality. Moreover, the kinetics of biomass and bio-compound production has been evaluated and the optimal physiological stage to collect plant material grown in open field has been identified. Biochemical characterization has been performed using the methods set up and collecting plant material in the optimal physiological stage identified in order to distinguish which genotypes were more suitable for bio-compound production purpose. The last focal point of PhD program was the development of an alternative technique for biomass and biocompound production in greenhouse grown conditions. Results obtained in the three PhD years, highlighted the possibility of using successfully some Cynara spp. genotypes for biomass and bio-compound production, in particular in open field condition. Also the real prospect of using some globe artichoke genotypes for food and non-food dual-production (biomass for biocompound extraction and heads for human food) has been underlined.

Freshwater bodies such as lakes release the greenhouse gas methane (CH4) into the atmosphere. Global emissions from lakes are estimated to emit more CH4 than oceans, despite that lakes occupies a much smaller global land area. Lakes are therefore significant components for global budgets of CH4. Accurate global estimations of lakes are troublesome, partly because of the spatial and temporal variability of CH4 fluxes, making regional and global assessments filled with uncertainties. Yet, few studies consider the spatial and temporal variability of CH4 fluxes. Therefore, this study investigates the spatial and temporal variability of CH4 fluxes in lakes at different scales. Measurements were made during two field campaigns in lake Venasjön and Parsen, located in the municipality of Söderköping, Sweden. We used the commonly used floating chamber (FC) method for CH4 flux measurements. In order to investigate the small-scale flux variability, we redeveloped the FC-method by constructing two grids consisting of seven FCs distributed approximately 1m apart from each other. One grid was placed at the shallow zone at the inflow of each lake and the other at the lakes deepest zone. By sampling the grid several times every field campaign, spatial and temporal variability of fluxes at different scales could be measured. Overall, we found a significant difference of CH4 fluxes in both lakes depending on field campaign and grid location. Our results also indicate that there is a small-scale variability of CH4 fluxes in lakes. Our hope is that these findings can illustrate the importance of investigating lake fluxes at small spatial and temporal scales. / Sjöar släpper ut växthusgasen metan (CH4) i atmosfären. Globala utsläpp från sjöar beräknas avge mer CH4 än havet, trots att sjöar har en mycket mindre global areal. Sjöar är därför viktiga komponenter för globala budgetar av CH4. Dessvärre är noggranna globala uppskattningar av sjöar svårt att göra, delvis på grund av den spatial och temporala variationen av CH4, vilket gör regionala och globala bedömningar fyllda med osäkerheter. Trots detta undersöker få studier metanflödets spatiala och temporala variabilitet. Denna studie undersöker den spatiala och temporala variabiliteten av CH4-flöden från sjöar i olika skalor. Mätningar genomfördes under två fältkampanjer i sjöarna Venasjön och Parsen som ligger i Söderköpings kommun, Sverige. För att undersöka variabiliteten i en liten skala utvecklade vi den redan använda floating chamber (FC) metoden för flödesmätningar genom att bygga två grids med sju FCs, ungefär 1m mellan varje kammare. En grid placerades vid den grunda delen vid inflödet av varje sjö och den andra vid respektive sjös djupaste del. Genom att mäta griden flera gånger varje fältkampanj kunde den spatiala och temporala variationen av flöden i olika skalor undersökas. Sammantaget fann vi signifikanta skillnader i båda sjöarnas CH4-flöden mellan fältkampanjer och grids. Våra resultat tyder också på småskaliga variationer av CH4-flöden i sjöar. Vårt hopp är att dessa resultat kan ytterligare bekräfta betydelsen av att undersöka sjöflödena i små spatiala och temporära skalor

Características de texturas atuam como bons descritores de imagens e podem ser empregadas em diversos problemas, como classificação e segmentação. Porém, quando o número de características é muito elevado, o reconhecimento de padrões pode ser prejudicado. A seleção de características contribui para a solução desse problema, podendo ser empregada tanto para redução da dimensionalidade como também para descobrir quais as melhores características de texturas para o tipo de imagem analisada. O objetivo deste trabalho é avaliar métodos ótimos e subótimos de seleção de características em problemas que envolvem texturas de imagens. Os algoritmos de seleção avaliados foram o branch and bound, a busca exaustiva e o sequential oating forward selection (SFFS). As funções critério empregadas na seleção foram a distância de Jeffries-Matusita e a taxa de acerto do classificador de distância mínima (CDM). As características de texturas empregadas nos experimentos foram obtidas com estatísticas de primeira ordem, matrizes de co-ocorrência e filtros de Gabor. Os experimentos realizados foram a classificação de regiôes de uma foto aérea de plantação de eucalipto, a segmentação não-supervisionada de mosaicos de texturas de Brodatz e a segmentação supervisionada de imagens médicas (MRI do cérebro). O branch and bound é um algoritmo ótimo e mais efiiente do que a busca exaustiva na maioria dos casos. Porém, continua sendo um algoritmo lento. Este trabalho apresenta uma nova estratégia para o branch and bound, nomeada floresta, que melhorou significativamente a eficiência do algoritmo. A avaliação dos métodos de seleção de características mostrou que os melhores subconjuntos foram aqueles obtidos com o uso da taxa de acerto do CDM. A busca exaustiva e o branch and bound, mesmo com a estratégia floresta, foram considerados inviáveis devido ao alto tempo de processamento nos casos em que o número de característica é muito grande. O SFFS apresentou os melhores resultados, pois, além de mais rápido, encontrou as soluções ótimas ou próximas das ótimas. Pôde-se concluir também que a precisão no reconhecimento de padrões aumenta com a redução do número de características e que os melhores subconjuntos freqüentemente são formados por características de texturas obtidas com técnicas diferentes / Texture features are eficient image descriptors and can be employed in a wide range of applications, such as classification and segmentation. However, when the number of features is considerably high, pattern recognition tasks may be compromised. Feature selection helps prevent this problem, as it can be used to reduce data dimensionality and reveal features which best characterise images under investigation. This work aims to evaluate optimal and suboptimal feature selection algorithms in the context of textural features extracted from images. Branch and bound, exhaustive search and sequential floating forward selection (SFFS) were the algorithms investigated. The criterion functions employed during selection were the Jeffries-Matusita (JM) distance and the minimum distance classifier (MDC) accuracy rate. Texture features were computed from first-order statistics, co-occurrence matrices and Gabor filters. Three different experiments have been conducted: classification of aerial picture of eucalyptus plantations, unsupervised segmentation of mosaics of Brodatz texture samples and supervised segmentation of MRI images of the brain. The branch and bound is an optimal algorithm and many times more eficient than exhaustive search. But is still time consuming. This work proposed a novel strategy for the branch and bound algorithm, named forest, which has considerably improved its performance. The evaluation of the feature selection methods has revealed that the best feature subsets were those computed by the MDC accuracy rate criterion function. Exhaustive search and branch and bound approaches have been considered unfeasible, due to their high processing times, especially for high dimensional data. This statement holds even for the branch and bound with the forest strategy. The SFFS approach yielded the best results. Not only was it faster, as it also was capable of finding the optimal or nearly optimal solutions. Finally, it has been observed that the precision of pattern recognition tasks increases as the number of features decreases and that the best feature subsets are those which possess features computed from distinct texture feature methods

Esta pesquisa investiga o comportamento sintático dos quantificadores flutuantes no português brasileiro (PB), dentro do modelo teórico do Programa Minimalista (Chomsky 1995) da Sintaxe Gerativa, buscando atingir dois objetivos principais: (i) identificar as posições na sentença em que os quantificadores podem ou não flutuar e (ii) identificar a estrutura sintagmática interna das expressões quantificadas. Com Sportiche (1988) assumimos que a flutuação de quantificadores é derivada por movimentos sintáticos e com Valmala Elguea (2008) assumimos que este fenômeno é relacionado ao conteúdo informacional da sentença. Dessa forma, adotamos a cartografia de projeções informacionais (de tópico e foco) proposta por Rizzi (1997) para a periferia esquerda alta da sentença e por Belletti (2004) para a periferia esquerda baixa (acima de VP). Argumentando que é preciso distinguir os tipos categoriais dos quantificadores para explicar por que alguns são capazes de flutuar no interior da sentença e outros não, mostramos que os quantificadores que projetam a categoria QP, como cada (um) e todos, são capazes de flutuar, enquanto quantificadores que projetam uma categoria igual ou menor que DP, como muitos, poucos, vários e alguns, são incapazes de flutuar no PB. Mostramos ainda como as diferenças estruturais internas entre todos (cujo DP associado é um complemento de Q0) e cada (um) (cujo DP associado é analisado como um adjunto de QP) podem explicar seu comportamento distinto em relação a algumas possibilidades de flutuação, considerando o cálculo de Minimalidade Relativizada entre QP e DP. A flutuação de quantificadores é | 10 | analisada como resultado da interação entre a estrutura sintagmática interna de cada expressão quantificada e outras propriedades da gramática (universais e específicas da língua) operantes ao longo da derivação e envolvidas sobretudo na checagem de Caso e na satisfação dos Critérios de Tópico e Foco. As possibilidades de flutuação devem respeitar a assimetria entre a periferia esquerda alta e a periferia esquerda baixa do PB em relação ao licenciamento de elementos com ou sem Caso sintático: enquanto DPs, NumPs e NPs que não foram licenciados na sintaxe podem ser superficializados com Caso default na periferia alta, a periferia baixa apenas pode superficializar DPs, NumPs e NPs licenciados antes de Spell-Out. Assumimos que quantificadores de categoria QP, além de poderem ter seu Caso licenciado na sintaxe por checagem, podem se licenciar na superfície por Caso default ou por Transmissão de Caso seja na periferia alta, na periferia baixa ou na posição temática propriedade que lhes garante a capacidade de flutuação. Propomos uma análise uniforme em que estruturas flutuantes e não-flutuantes podem ser geradas a partir de um mesmo constituinte subjacente inserido na posição temática, sendo distintas por seu percurso derivacional. Dessa forma, esta análise simplifica e uniformiza algumas questões presentes na literatura sobre a flutuação de quantificadores. / This research investigates the syntactic behavior of floating quantifiers in Brazilian Portuguese (BP), in the framework of the Minimalist Program (Chomsky 1995), aiming at two main goals: (i) to identify the sentence positions in which quantifiers can or cannot float and (ii) to identify the internal phrase structure of quantified expressions. I assume with Sportiche (1988) that quantifier floating is derived through syntactic movement and with Valmala Elguea (2008) that this phenomenon is related to the informational content of the sentence. Thus, I adopt the cartography of informational projections (topic and focus) proposed by Rizzi (1997) for the high left periphery of the sentence and by Belletti (2004) for the low left periphery (above VP). I argue that in order to explain why some quantifiers are able to float inside the sentence and some are not, it is necessary to distinguish the categorial types of quantifiers. I show that quantifiers that project as QPs, like cada (um) each (one) and todos all, are able to float, whereas quantifiers that project as DPs or lower categories, like muitos many, poucos few, vários several and alguns some, are unable to float in BP. We also show how internal structural differences between todos (whose associate DP is a complement of Q0) and cada (um) (whose associate DP is analyzed as an adjunct of QP) can explain their different behavior regarding some floating possibilities, given the computation of Relativized Minimality between QP and DP. Quantifier floating is analyzed here as the result of the interaction between the internal phrase structure of each quantified expression and other properties of the grammar (both universal and language| 12 | specific), in particular Case-checking and Topic and Focus Criteria. Floating possibilities must respect the asymmetry between the high and the low left peripheries in BP regarding the licensing of elements with or without syntactic Case: whereas DPs, NumPs and NPs that have not been licensed in the syntax can surface with default Case in the high periphery, the low periphery can only host DPs, NumPs and NPs that have already been licensed before Spell- Out. I assume that QP-type quantifiers, besides being able to have their Case licensed in the syntax via Checking, can be licensed on the surface via default Case or Case Transmission regardless of whether they are in the high left periphery, in the low left periphery or in the thematic position which is in fact the property that provides them with the ability to float. We propose a uniform analysis in which both floated and non-floated structures can be generated from the same underlying constituent inserted in the thematic position, being distinguished by their derivational course. Therefore, this analysis simplifies and unifies some issues present in the literature on quantifier floating.

Este trabalho apresenta o estudo do efeito de elevação atípica da transcondutância na região linear de polarização devido ao efeito de corpo flutuante induzido pela porta (Gate Induced Floating Body Effect - GIFBE) de transistores da tecnologia SOI nMOSFET. Este estudo foi realizado com base em resultados experimentais e em simulações numéricas, as quais foram essenciais para o entendimento físico deste fenômeno. Além de contribuir com a explicação física deste fenômeno, este trabalho explora o efeito de corpo flutuante em diferentes estruturas (transistor de porta única, transistor de porta gêmea, transistor de múltiplas portas e transistores de canal tensionado), diferentes tecnologias e em função da temperatura (100K a 450K). A partir do estudo realizado em dispositivos SOI de porta única analisouse a influência das componentes da corrente de porta que tunelam através do óxido de porta do dispositivo, o potencial da região neutra do corpo do transistor, a taxa de recombinação de portadores, o impacto da redução da espessura do óxido de porta e também as dimensões físicas do transistor. Na análise feita da redução do comprimento de canal, verificou-se também que o GIFBE tende a ser menos significativo para dispositivos ultra-submicrométricos. Analisou-se também o efeito da elevação atípica da transcondutância para transistores SOI totalmente depletados, para os quais, este efeito ocorre apenas quando a segunda interface está acumulada, para as duas tecnologias estudadas (65nm e 130nm). A análise dos dispositivos de porta gêmea, que tradicionalmente são usados com a finalidade de minimizar o efeito de elevação abrupta de corrente de dreno, mostrou uma redução do GIFBE para este tipo de estrutura quando comparada à de porta única devido ao aumento da resistência série intrínseca à estrutura. O efeito de corpo flutuante também foi avaliado em função da temperatura de operação dos dispositivos. Para temperaturas variando de 100K a 450K, notou-se que o valor do limiar de GIFBE aumentou tanto para temperaturas acima de 300K quanto abaixo da mesma. Quando estes resultados são apresentados graficamente, observa-se que o comportamento do limiar de GIFBE com a temperatura resulta no formato de uma letra C, onde o valor mínimo está a 300K. Este comportamento se deve à competição entre o processo de recombinação e a degradação efetiva da mobilidade. Uma primeira análise do GIFBE em diferentes estruturas de transistores também foi realizada. Apesar dos transistores de canal tensionado apresentarem o efeito para valores menores de tensão de porta, este efeito se manifesta com menor intensidade nestes transistores, devido a alta degradação da mobilidade efetiva apresentada pelo mesmo. Entretanto, quando o foco são os transistores de múltiplas portas, os resultados obtidos demonstram que apesar destes dispositivos terem sido fabricados com dielétrico de porta de alta constante dielétrica, o GIFBE ainda ocorre. Esta ocorrência do GIFBE em FinFETs é fortemente dependente da largura do Fin, da dopagem da região de canal e conseqüentemente do acoplamento das portas laterais com a superior. / This work presents the study of the Gate Induced Floating Body Effect (GIFBE) that occurs in the SOI MOSFET technology. This study has been performed based on experimental results and on numerical simulations, which were an essential auxiliary tool to obtain a physical insight of this effect. Besides the contribution on the physical explanation of this phenomenon, in this work, the floating body effect was evaluated for different structures (single gate and twin-gate transistors), different technologies (130nm and 65nm SOI CMOS technology) and as a function of the temperature (100K to 450K). From the study of the single gate devices, it was evaluated the gate tunneling current influence on GIFBE, the body potential in the neutral region, the recombination rate, the front gate oxide thickness reduction impact, besides the physical dimensions of the transistor. In the performed analysis, taking into account the channel length reduction, it was verified that the GIFBE tends to be less important for ultra-submicron devices. The GIFBE only occurs for fully depleted devices when the second interface is accumulated. In this situation, the floating body effect influence on fully depleted devices was also studied for both technologies (65nm and 130nm). The twin-gate devices analysis, that traditionally are used in order to minimize the Kink effect, show a GIFBE reduction for this structure when it is compared to the single gate one. This enhance in the electrical characteristics is due to the series resistance increase that is intrinsic of this structures. When the temperature variation from 100K to 450K was analyzed, it was obtained the C shape behavior for the floating body effect due to a competition between the recombination process and the effective mobility degradation factor. A first evaluation of the GIFBE occurrence in new devices was also performed. When the focus is the strained silicon transistor, a occurrence of GIFBE was obtained for a lower gate voltage. Although, the GIFBE occurs earlier for strained transistor. This effect is less pronounced in this device because it presents strong effective mobility degradation. When the focus is FinFETs, the results show that although this device was fabricated with a high-k gate dielectric, the GIFBE still occurs and is strongly dependent on the device channel width.

The paper compares electricity production between the semi-submersible and the spar-buoy floating wind turbine systems under normal, stochastic and extreme wind conditions at Utsira Nord site located on the Norwegian continental shelf in the North Sea. The analysis of complex behavior of the floating wind turbine system and the fluid-structure interaction is performed in aero-servo-hydro-elastic code ASHES. The results indicate a slightly better energy performance of the semi-submersible than the spar in all load cases but one. The pitch and heave degrees of freedom are evaluated as the most relevant for the power output. It is shown that pitch and heave platform motions have smaller displacement in the semi-submersible floater than in the spar under average environmental conditions and at the rated wind speed operating range. The simulation also confirmed that the energy yield is very sensitive to the magnitude of the loads: the spar performed best under mild environmental conditions, while the semi-submersible was better under medium environmental conditions. Small difference in energy yield is attributed to the same baseline blade and external controller properties used for both floaters where generator torque was kept constant to limit the power excursions above the rated power. The method proposed under this paper has demonstrated that a good approximation of the energy performance of the floating wind turbine system can be performed in a fast and effective manner.

Having gained popularity in the last decade, the dielectric barrier discharge (DBD) has been studied in numerous ways as a device for air acceleration. A major interest of all these works has been concentrated on achieving higher efficiency and producing higher jet velocities. These studies considered alternative geometries, different voltage waveforms etc., yet none of these studies tackle the inherently inefficient process of force creation and actuation in a DBD. Air acceleration has been attributed to the electric force that switches direction on each voltage half cycle (that is due to the nature of the discharge mechanism) yet has a forward resulting force. The present thesis considers the phenomenon of memory charge accumulation on the dielectric surface and reasons that the backward force in the second half cycle can be reversed by further migration of the memory charges to an added exposed anode. Hence a novel 3-electrode configuration is presented. Flow velocity investigations showed a 27% improvement in efficiency in comparison with the traditional 2-electrode actuator. This meant a 0.4m/s gain along the velocity profile. During the investigation process, a new discharge mechanism was encountered. This was characterized by the double discharge lines along the exposed electrode edges while in the presence of a \"floating\" encapsulated electrode. / Tendo ganho popularidade na última década, a descarga de barreira dielétrica (DBD) foi estudada de várias maneiras como um dispositivo para aceleração do ar e controle aerodinâmico. Um grande interesse de todos esses trabalhos foi concentrado em alcançar maior eficiência e produzir maiores velocidades de jato. Alguns desses estudos consideraram geometrias alternativas, diferentes formas de onda de tensão, etc., mas nenhum desses estudos aborda o processo inerentemente ineficiente de criação e atuação da força em um DBD. A Aceleração do ar tem sido atribuída à força elétrica que muda de direção em cada meio ciclo (que é devido à natureza do mecanismo de descarga), mas que tem uma força resultante para a frente. A presente tese considera o fenômeno de acumulação de carga de memória na superfície dielétrica e raciocina que a força contraria no segundo meio ciclo da descarga pode ser revertida por migração dessas cargas de memória para um ânodo adicional exposto. Assim, uma nova configuração de 3 eletrodos é apresentada. Investigações de velocidade do fluxo mostraram uma melhoria de 27% na eficiência em comparação com o tradicional atuador de 2 eletrodos. Isso significou um ganho de 0,3 m/s ao longo do perfil de velocidade. Durante o processo de investigação, foi encontrado um novo mecanismo de descarga. Isso foi caracterizado por uma dupla descarga ao longo das bordas do eletrodos expostos, enquanto na presença de um eletrodo encapsulado \"pendurado\".

Une sous-classe intéressante de matériaux multiferroïques est celle des composés multiferroïques magnétoélectriques, dans lesquels il existe un couplage entre les paramètres d’ordres ferroïques (magnétique et électrique). De ce point de vue, la classe des matériaux Ga₂₋ₓFeₓO₃ a attiré l’attention des chercheurs. Ces composés sont actuellement connus pour leur température de transition élevée ainsi que pour l’interaction possible entre leurs propriétés ferrimagnétiques et piézoélectriques. Leur structure cristallographique et magnétique est assez complexe, du fait du désordre de substitution interne Fe/Ga. Les oxydes M₂Ga₂Fe₂O₉ (M=In, Sc) appartiennent à cette même famille de matériaux et ont été synthétisés pour la première fois afin d’obtenir une structure cristallographique ordonnée de GaFeO₃. Afin d’étudier les propriétés physiques de ces différents composés, nous avons synthétisé par la méthode de la zone flottante (au four à image), en utilisant différentes conditions de croissance, des monocristaux de composition Ga₂₋ₓFeₓO₃ (x=0.90, 1.00 et 1.10). Nous avons également élaboré des échantillons polycristallins de composés GaFeO₃ faiblement dopés en indium ainsi que le composé M₂Ga₂Fe₂O₉ (M=In, Sc). Nous avons enfin préparé de monocristaux de composition In₂Ga₂Fe₂O₉ par la méthode de croissance en flux. L’affinement Rietveld des diffractogrammes des rayons X et des neutrons nous a permis de montrer que les céramiques de GaFeO₃ faiblement dopées en indium et les monocristaux de Ga₂₋ₓFeₓO₃ cristallisent dans le groupe d’espace Pc2₁n. Les paramètres cristallins et la température de Néel caractéristiques pour les monocristaux de Ga₂₋ₓFeₓO₃ varient de manière linéaire avec la teneur en fer. Les affinements nous ont permis de conclure que la structure de ces composés est caractérisée par un désordre élevée (25% de la quantité du fer se trouve sur les sites natifs du gallium). L’incorporation graduelle de l’indium s’accompagne d’une augmentation du volume de la maille ainsi qu’à une diminution de la température de transition magnétique. Le spectre d’excitations magnétiques mesuré pour les cristaux de Ga₂₋ₓFeₓO₃ nous a permis de mettre en évidence une coexistence de l’ordre ferrimagnétique à longue portée et d’un signal de diffusion diffuse en-dessous de la température de Néel. Ce signal diffus suggère l’existence d’une composante de type verre de spin du fait du désordre interne des sites. L'étude de la variation thermique de la constante diélectrique sur un cristal de GaFeO₃ révèle l’absence d’un couplage magnétoélectrique au sein de ces matériaux. L’affinement Rietveld des diagrammes de diffraction des rayons X et des neutrons mesurés sur les poudres de M₂Ga₂Fe₂O₉ (M=In, Sc) révèle une structure orthorhombique de type Pba2 fortement désordonnée, avec quatre sites cationiques d’occupation mixte. Les données de susceptibilité DC et AC couplées avec les mesures de chaleur spécifique et les spectres Mössbauer indiquent, en-dessous d’une température de Tg ≈ 19 K, l’existence d’un état fondamental de type verre de spin dans ce système. Les mesures du spectre d’excitations magnétiques ont mis en évidence l’absence d’ordre magnétique à longue portée et confirment l’existence d’une transition d’un état paramagnétique vers un état verre de spins. L’existence d’un comportement de type verre de spin dans les systèmes Ga₂₋ₓFeₓO₃ et M₂Ga₂Fe₂O₉ (M=In, Sc) souligne l’importance du désordre interne pour la caractérisation de l'état fondamental magnétique. / An interesting sub-class of the multiferroic materials are the multiferroic magnetoelectrics, in which exists a coupling between the two ferroic order parameters (magnetic and electric). From this viewpoint, the case of the (Ga₂₋ₓFeₓO₃)-class of materials has retained special attention. These compounds are now well-known for their high transition temperature as well as for the potential interaction between their ferrimagnetic and piezoelectric properties. Their crystallographic and magnetic structure are quite complicated, due to the existence of an internal site disorder. M₂Ga₂Fe₂O₉ (M=In, Sc) belongs to the (Ga₂₋ₓFeₓO₃)-class of materials and it was first synthesized in an attempt to obtain an ordered GaFeO₃ crystallographic structure. In order to study the physical properties of these compounds, Ga₂₋ₓFeₓO₃ (x=0.90, 1.00 and 1.10) single crystals have been synthesized by the floating zone method in an infrared image furnace, using different growth conditions. Indium doped GaFeO₃ (up to 10% indium content amount) and M₂Ga₂Fe₂O₉ (M=In, Sc) polycrystalline materials have been prepared by solid state reaction. Also, In₂Ga₂Fe₂O₉ single crystals were prepared by the flux method. The indium doped GaFeO₃ and Ga₂₋ₓFeₓO₃ samples crystallize in the Pc2₁n space group as determined from Rietveld refinement of the X-ray and neutron single crystals and powder patterns. The cell parameters and the Néel temperature (TN) of the Ga₂₋ₓFeₓO₃ single crystals varies linearly with the iron content amount. The occupation factors were calculated by refinement and the results showed a disordered structure (25% of the iron amount is found on the native gallium sites). The gradual incorporation of indium is accompanied by an increase of the cell volume and a decrease of the magnetic transition temperature. The magnetic excitations spectra measured for the Ga₂₋ₓFeₓO₃ single crystals revealed a coexistence of a ferrimagnetic ordering and a diffuse scattering signal below the Néel temperature. The diffuse signal suggests the existence of a spin glass like component due to the internal site disorder. Dielectric investigations showed no temperature dependent anomaly of the dielectric constant for the GaFeO₃ single crystal, suggesting a lack of a magnetoelectric coupling signal in this system. The Rietveld refinement of the X-ray and neutron powder patterns for the M₂Ga₂Fe₂O₉ (M=In, Sc) revealed a highly disordered orthorhombic Pba2 structure, offering four mixed cationic crystallographic sites. DC and AC susceptibility data in conjunction with the heat capacity data and Mössbauer measurements indicated a spin-glass-like behavior in this system, with a freezing temperature near Tf ≈ 19 K. The absence of an long range magnetic ordering and the existence of a transition from a paramagnetic state to a «frozen» state were showcased by the spin dynamics spectra measurements. The evidence for glassy behavior in the Ga₂₋ₓFeₓO₃ and M₂Ga₂Fe₂O₉ (M=In, Sc) systems highlights the importance of site disorder in determining the ground state magnetic properties.

CoordenaÃÃo de AperfeiÃoamento de Pessoal de NÃvel Superior / Visando superar os desafios inerentes à conversÃo de energia elÃtrica em sistemas de alta potÃncia, minimizando as perdas e melhorando a qualidade da energia processada, este tra-balho tem por objetivo analisar e implementar uma tÃcnica de modulaÃÃo para ser aplicada nas duas topologias de inversores multinÃveis mais disseminadas - com neutro grampeado (NPC), e com capacitor flutuante (FC) - a fim de reduzir os esforÃos nos semicondutores, bem como melhorar o Ãndice de distorÃÃo harmÃnica da tensÃo de saÃda. Ao longo do trabalho foi evidenciada a necessidade da digitalizaÃÃo da tÃcnica proposta, visto que o desenvolvimento analÃgico da mesma acarretaria em um circuito complexo e de baixa confiabilidade. Dessa forma, escolheu-se como plataforma digital um FPGA, devido à sua facilidade de programa-ÃÃo e reconfiguraÃÃo, alÃm da alta velocidade e quantidade de pinos de entrada e saÃda. AlÃm da tÃcnica proposta, foram tambÃm desenvolvidas outras modulaÃÃes para fins de compara-ÃÃo, apresentando os padrÃes de chaveamento para cada uma delas, bem como o comporta-mento da corrente atravÃs dos semicondutores em cada perÃodo de chaveamento. Foi tambÃm realizada a anÃlise teÃrica das topologias e suas respectivas etapas de operaÃÃo, caracterÃsticas e levantamento das equaÃÃes que ditam a anÃlise das perdas para as diferentes situaÃÃes de tÃcnicas aplicadas a cada uma das estruturas. O desenvolvimento digital das tÃcnicas mostrou-se correta atravÃs da anÃlise das formas-de-onda colhidas por meio de um circuito digital-analÃgico. AlÃm disso, a comparaÃÃo da aplicaÃÃo dessas modulaÃÃes em inversores a trÃs nÃveis NPC e FC de 6 kW mostrou-se favorÃvel à tÃcnica proposta em termos de eficiÃncia e reduÃÃo da distorÃÃo harmÃnica em ambas as topologias, comprovando sua utilidade em con-versores multinÃveis de alta potÃncia. Por fim, foi apresentado o desenvolvimento da tÃcnica proposta em inversores com mais de trÃs nÃveis, onde se pode comprovar sua eficiente aplica-ÃÃo para tais fins, bem como sua expansibilidade para inversores de n nÃveis. / In order to overcome the challenge of processing electric energy in high power systems with minimal losses and high energy quality, this work presents the implementation and anal-ysis of a modulation technique applicable on both most well-known multilevel inverter struc-tures - neutral point-clamped (NPC), and flying capacitors (FC) - to reduce the stresses across the semiconductors devices, and to improve the total harmonic distortion of the output volt-age. Throughout the work, the necessity to digitalize the proposed technique has been evi-denced due to the high complexity and low reliability inherent to the analogical approach. Thus, the digital controller FPGA has been chosen, as it is easy to program and reconfigure, works at high speed, and has a lot of input and output pins. Additionally, other modulation techniques were also implemented to compare their performance with the proposed one, pre-senting the switching patterns and the behavior of the electrical currents through the semicon-ductors for each modulation. A theoretical analysis was also performed for both topologies and their respective operation principle, characteristics, and equations used on the losses anal-ysis for the different combinations of modulation applied to each structure. Finally, the digital development of the various techniques has proved to be correct by observing the waveforms obtained through the digital/analogical circuit. Besides, the comparison of the modulation techniques on 6 kW NPC and FC three-level prototype inverters proved to be favorable to the proposed technique in terms of efficiency and total harmonic distortion reduction on both topologies, confirming its usefulness on high power multilevel converters. At last, it was pre-sented the application of the proposed modulation technique to inverters with more than three levels, where it was observed its eligibility for n-levels topologies.

Características de texturas atuam como bons descritores de imagens e podem ser empregadas em diversos problemas, como classificação e segmentação. Porém, quando o número de características é muito elevado, o reconhecimento de padrões pode ser prejudicado. A seleção de características contribui para a solução desse problema, podendo ser empregada tanto para redução da dimensionalidade como também para descobrir quais as melhores características de texturas para o tipo de imagem analisada. O objetivo deste trabalho é avaliar métodos ótimos e subótimos de seleção de características em problemas que envolvem texturas de imagens. Os algoritmos de seleção avaliados foram o branch and bound, a busca exaustiva e o sequential oating forward selection (SFFS). As funções critério empregadas na seleção foram a distância de Jeffries-Matusita e a taxa de acerto do classificador de distância mínima (CDM). As características de texturas empregadas nos experimentos foram obtidas com estatísticas de primeira ordem, matrizes de co-ocorrência e filtros de Gabor. Os experimentos realizados foram a classificação de regiôes de uma foto aérea de plantação de eucalipto, a segmentação não-supervisionada de mosaicos de texturas de Brodatz e a segmentação supervisionada de imagens médicas (MRI do cérebro). O branch and bound é um algoritmo ótimo e mais efiiente do que a busca exaustiva na maioria dos casos. Porém, continua sendo um algoritmo lento. Este trabalho apresenta uma nova estratégia para o branch and bound, nomeada floresta, que melhorou significativamente a eficiência do algoritmo. A avaliação dos métodos de seleção de características mostrou que os melhores subconjuntos foram aqueles obtidos com o uso da taxa de acerto do CDM. A busca exaustiva e o branch and bound, mesmo com a estratégia floresta, foram considerados inviáveis devido ao alto tempo de processamento nos casos em que o número de característica é muito grande. O SFFS apresentou os melhores resultados, pois, além de mais rápido, encontrou as soluções ótimas ou próximas das ótimas. Pôde-se concluir também que a precisão no reconhecimento de padrões aumenta com a redução do número de características e que os melhores subconjuntos freqüentemente são formados por características de texturas obtidas com técnicas diferentes / Texture features are eficient image descriptors and can be employed in a wide range of applications, such as classification and segmentation. However, when the number of features is considerably high, pattern recognition tasks may be compromised. Feature selection helps prevent this problem, as it can be used to reduce data dimensionality and reveal features which best characterise images under investigation. This work aims to evaluate optimal and suboptimal feature selection algorithms in the context of textural features extracted from images. Branch and bound, exhaustive search and sequential floating forward selection (SFFS) were the algorithms investigated. The criterion functions employed during selection were the Jeffries-Matusita (JM) distance and the minimum distance classifier (MDC) accuracy rate. Texture features were computed from first-order statistics, co-occurrence matrices and Gabor filters. Three different experiments have been conducted: classification of aerial picture of eucalyptus plantations, unsupervised segmentation of mosaics of Brodatz texture samples and supervised segmentation of MRI images of the brain. The branch and bound is an optimal algorithm and many times more eficient than exhaustive search. But is still time consuming. This work proposed a novel strategy for the branch and bound algorithm, named forest, which has considerably improved its performance. The evaluation of the feature selection methods has revealed that the best feature subsets were those computed by the MDC accuracy rate criterion function. Exhaustive search and branch and bound approaches have been considered unfeasible, due to their high processing times, especially for high dimensional data. This statement holds even for the branch and bound with the forest strategy. The SFFS approach yielded the best results. Not only was it faster, as it also was capable of finding the optimal or nearly optimal solutions. Finally, it has been observed that the precision of pattern recognition tasks increases as the number of features decreases and that the best feature subsets are those which possess features computed from distinct texture feature methods