Made available in DSpace on 2014-06-12T15:58:28Z (GMT). No. of bitstreams: 2 arquivo4572_1.pdf: 1149011 bytes, checksum: 92aae8f6f9b5145bfcecb94d96dbbc0b (MD5) license.txt: 1748 bytes, checksum: 8a4605be74aa9ea9d79846c1fba20a33 (MD5) Previous issue date: 2004 / Instituto Federal de Educação, Ciência e Tecnologia de Pernambuco / Um nariz artificial é um sistema modular composto de duas partes principais: um sistema sensor, formado de elementos que detectam odores e um sistema de reconhecimento de padrões que classifica os odores detectados. Redes neurais artificiais têm sido utilizadas como sistema de reconhecimento de padrões para narizes artificiais e vêm apresentando resultados promissores. Desde os anos 80, pesquisas para criação de narizes artificiais, que permitam detectar e classificar odores, vapores e gases automaticamente, têm tido avanços significativos. Esses equipamentos podem ser utilizados no monitoramento ambiental para controlar a qualidade do ar, na área de saúde para realizar diagnóstico de doenças e nas indústrias de alimentos para o controle de qualidade e o monitoramento de processos de produção. Esta dissertação investiga a utilização de quatro técnicas diferentes de redes neurais para criação de sistemas de reconhecimento de padrões em narizes artificiais. O trabalho está dividido em quatro partes principais: (1) introdução aos narizes artificiais, (2) redes neurais artificiais para sistema de reconhecimento de padrões, (3) métodos para medir o desempenho de sistemas de reconhecimento de padrões e comparar os resultados e (4) estudo de caso. Os dados utilizados para o estudo de caso, foram obtidos por um protótipo de nariz artificial composto por um arranjo de oito sensores de polímeros condutores, expostos a nove tipos diferentes de aguarrás. Foram adotadas as técnicas Multi-Layer Perceptron (MLP), Radial Base Function (RBF), Probabilistic Neural Network (PNN) e Time Delay Neural Network (TDNN) para criar os sistemas de reconhecimento de padrões. A técnica PNN foi investigada em detalhes, por dois motivos principais: esta técnica é indicada para realização de tarefas de classificação e seu treinamento é feito em apenas um passo, o que torna a etapa de criação dessas redes muito rápida. Os resultados foram comparados através dos valores dos erros médios de classificação utilizando o método estatístico de Teste de Hipóteses. As redes PNN correspondem a uma nova abordagem para criação de sistemas de reconhecimento de padrões de odor. Estas redes tiveram um erro médio de classificação de 1.1574% no conjunto de teste. Este foi o menor erro obtido entre todos os sistemas criados, entretanto mesmo com o menor erro médio de classificação, os testes de hipóteses mostraram que os classificadores criados com PNN não eram melhores do que os classificadores criados com a arquitetura RBF, que obtiveram um erro médio de classificação de 1.3889%. A grande vantagem de criar classificadores com a arquitetura PNN foi o pequeno tempo de treinamento dos mesmos, chegando a ser quase imediato. Porém a quantidade de nodos na camada escondida foi muito grande, o que pode ser um problema, caso o sistema criado deva ser utilizado em equipamentos com poucos recursos computacionais. Outra vantagem de criar classificadores com redes PNN é relativa à quantidade reduzida de parâmetros que devem ser analisados, neste caso apenas o parâmetro relativo à largura da função Gaussiana precisou ser investigado

La plaine de la Crau (Bouches-du-Rhône, France) renferme une nappe phréatique alluviale qualifiée de « ressource patrimoniale ». Débutée il y a près de 500 ans par la création d’un réseau de canaux, la mise en culture de prairies, est encore aujourd’hui, pratiquée selon une technique d’irrigation traditionnelle par submersion. Provenant d’un autre bassin versant, les eaux d’irrigation ont un effet majeur sur la recharge de la nappe. Néanmoins l’urbanisation progressive du territoire, l’augmentation des prélèvements, et le changement climatique, menacent l’équilibre actuel de la nappe de la Crau. La préservation de cette ressource nécessite donc la production d’un outil de gestion performant prenant en compte la globalité de cet hydro-système, ainsi que sa complexité géométrique et hydrologique. Cette thèse présente une approche originale de modélisation hydrogéologique en estimant à chaque étape les variables et paramètres nécessaires de manière indépendante, réduisant ainsi un problème récurrent d’équifinalité. Constituant donc un outil prospectif fiable, ce nouveau modèle est alimenté par des scénarios réalistes permettant d’observer les impacts du changement climatique, de l’évolution de l’occupation des sols planifiée et de réductions occasionnelles de la dotation en eaux d’irrigation à l’horizon 2030. Sous les effets combinés de ces réductions de la recharge (jusqu’à -19%), la nappe phréatique de la Crau pourrait être soumise à une diminution de sa surface piézométrique allant jusqu’à 2 m entrainant l’assèchement de zones humides rares et protégées. / The Crau plain (Southern France) contains an alluvial aquifer described as a regional "heritage resource". Started nearly 500 years ago by creating a network of canals, the cultivation of grasslands is even today performed using a traditional technique of irrigation by flooding. Derived from another watershed, irrigation water has a major impact on the recharge of the aquifer. Nevertheless, urbanization of the territory, increase of uptakes, and climate change threaten the existing balance of the Crau aquifer. The preservation of this resource therefore requires the production of an efficient management tool accounting for the whole hydro-system in its hydrological and geometric complexity. This thesis presents an original approach of hydrogeological modeling by independently estimating, for each step, the necessary variables and parameters, reducing a recurring problem of equifinality. Providing a reliable forecasting tool, this new model is implemented by realistic scenarios to observe the impacts of climate change, the evolution of the planned land-use, and occasional reductions of irrigation input in 2030. Under the combined effects of these recharge reductions (up to -19%) the water table could be subjected to local decreases up to 2 m, leading to the drying up of rare and protected wetlands.

L’usage croissant des matériaux composites dans l’industrie induit de nouvelles problématiques dans des domaines variés,notamment pour la caractérisation non destructive. Les méthodes courantes comme l’analyse modale ou les éléments finis sont rarement adaptées pour représenter la dynamique vibratoire complexe des structures composites ou quantifier leurs caractéristiques viscoélastiques, de nouvelles approches sont nécessaires.Les travaux concernent le développement et l’application d’un formalisme vibratoire inverse local, la méthode RIFF(Résolution Inverse Filtrée Fenêtrée), pour l’étude des matériaux composites multicouches. Ces derniers sont considérés comme homogènes à l’aide de la théorie de Timoshenko, prenanten compte le cisaillement non négligeable de ces matériaux.Une caractérisation fréquentielle et/ou spatiale des paramètres équivalents (modules d’Young E et de cisaillement G, et facteurs de pertes associés) est alors possible, permettant de traduire fidèlement le comportement dynamique des composites et de simplifier leur modélisation en éléments finis.Une seconde approche utilisant un schéma aux différences finies corrigé (méthode RIC) autorise une analyse similaire à partir d’un maillage grossier diminuant fortement les temps de mesure et de post-traitement des données.Enfin, une perspective de détection et d’identification de défauts est envisagée. Grâce à des cartographies des paramètres élastiques et d’amortissements, il semble possible de pouvoir déduire la signature d’un défaut typique. Une discontinuité du module de cisaillement témoignerait de la présence d’un délaminage, la diminution du module d’Young traduirait une rupture de fibres, etc. / The increasing use of composite materials in the industry leadsto new challenges in various areas, including non-destructiveevaluation. Common methods such as modal analysis or finiteelements are rarely appropriated to represent the complexvibratory dynamic of composite structures or quantify theirviscoelastic properties, new approaches are then needed.This thesis deals with the development and application of a localinverse vibratory method, called the Force Analysis Technique(FAT), in order to the study multilayer composites. The latterare considered to be homogeneous using the Timoshenko beamtheory, which takes shear effects into account, usually significantfor such structures. A frequency and/or spatial characterizationof the equivalent elastic parameters (Young’s modulus E, shearmodulus G and their associated loss factors) isthen possible to accurately interpret the dynamical behaviourof composite materials and also simplify their implementationin finite element software.A second approach using a corrected finite difference scheme(CFAT method) allows a similar analysis using a coarse mesh,reducing the durations of measurement and post-processing.Finally, a perspective of detection and identification of defects isconsidered. By mean of cartographies of the elastic parameters,it seems possible to infer a signature related to a kind of flaw. Adiscontinuity of the shear modulus would attest the presence ofdelamination while a reduced Young’s modulus could indicate afibre breakage, etc.

Several geotechnical processes can only be reliably interpreted by taking account of the soil-atmosphere interactions. This thesis investigates two geotechnical problems involving soil-atmosphere interactions that drive water flow through the soil skeleton in two opposite directions; Problem 1: slope failure in expansive soils induced by water infiltration, Problem 2: large strain consolidation of soft soils induced by water evaporation. Both problems are of practical interest for safe and economical design of various geotechnical infrastructures. Two major geotechnical activities in the world; namely, the construction of water transfer canal in expansive soil area in China and the deposition of oil sands and hard rock tailings in Canada can be cited as classic examples of Problems 1 and 2, respectively. In such problems, substantial zones of the domain may switch between an unsaturated and saturated condition. Therefore, rational analysis requires simultaneous modelling of both unsaturated and saturated soil behaviour. The first goal of this thesis is to investigate the influence of swelling (the most characteristic behaviour of expansive soils) on slope stability using numerical methods. Swelling of expansive soils contributes to slope instability during rainfall because of two key reasons (i) soil swelling affects the flow process that actually induces swelling, (i.e. a typical coupling phenomenon), and (ii) swelling-induced stress redistribution and displacement development. In this thesis, the first effect is studied by a coupled (mechanical-hydraulic) numerical analysis of the response of a slope to rainfall using commercial software (GeoSlope). The second effect, the swelling-induced stress redistribution and displacement development after wetting, is tracked using a newly developed numerical program. In the program strain softening behaviour is introduced into the elasto-plastic Mohr-Coulomb Model for modelling unsaturated soil. A novel stress (net stress and suction)-dependent model for moduli of elasticity, combined with the predictive model for shear strength based on Soil Water Retention Behaviour are incorporated into the numerical program to achieve a smooth transition between saturated and unsaturated states. The results show that soil swelling can decrease the factor of safety by accelerating the wetting front depth due to hydro-mechanical coupling, while changes of sliding mass geometry has a negligible influence. The change of stress regime associated with soil swelling is significant to induce plastic strain softening (swelling-induced softening) and contribute to the slope failures. The second goal of thesis is to develop a novel computer program for simulation of large strain consolidation of soft soil under both self-weight and evaporation conditions. This program is both theoretically sound and practically applicable. Several basic/advanced constitutive models for unsaturated soils, including State Surface Model (SSM), Barcelona Basic model (BBM), Glasgow Coupled model (GCM) and bounding surface water retention model, are innovatively implemented into a piece-wise linear framework solved using finite difference technique. The developed program is referred to as UNSATCON-(ML), which has been tested using (a) existing analytical/numerical solutions and (b) various laboratory and field studies for single-layer and multiple-layer deposition of hard rock and oil sands tailings. Features of UNSATCON-(ML) that are improvements over existing models typically used to analyze consolidation-desiccation in soft soils include: (i) coupling of soil large deformation with true unsaturated water flow; (ii) correct reproduction of the shrinkage behaviour of soil under evaporation-induced desiccation; (iii) smooth transition between saturated and unsaturated states despite that some selected models are established using two independent stress variables, (iv) ensuring strictly mass conservation of water, and (v) simulation of irrecoverable volume change and hydraulic hysteresis to properly analyze multilayer tailings deposition. A number of hypothetical field case analyses are carried out using UNSATCON-ML, illustrating its applicability to industry.

This research investigates the use of an alternate means of modelling the performance of a train of feed water heaters in a steam cycle power plant, using machine learning. The goal of this study was to use a simple artificial neural network (ANN) to predict the behaviour of the plant system, specifically the inlet bled steam (BS) mass flow rate and the outlet water temperature of each feedwater heater. The output of the model was validated through the use of a thermofluid engineering model built for the same plant. Another goal was to assess the ability of both the thermofluid model and ANN model to predict plant behaviour under out of normal operating circumstances. The thermofluid engineering model was built on FLOWNEX® SE using existing custom components for the various heat exchangers. The model was then tuned to current plant conditions by catering for plant degradation and maintenance effects. The artificial neural network was of a multi-layer perceptron (MLP) type, using the rectified linear unit (ReLU) activation function, mean squared error (MSE) loss function and adaptive moments (Adam) optimiser. It was constructed using Python programming language. The ANN model was trained using the same data as the FLOWNEX® SE model. Multiple architectures were tested resulting in the optimum model having two layers, 200 nodes or neurons in each layer with a batch size of 500, running over 100 epochs. This configuration attained a training accuracy of 0.9975 and validation accuracy of 0.9975. When used on a test set and to predict plant performance, it achieved a MSE of 0.23 and 0.45 respectively. Under normal operating conditions (six cases tested) the ANN model performed better than the FLOWNEX® SE model when compared to actual plant behaviour. Under out of normal conditions (four cases tested), the FLOWNEX SE® model performed better than the ANN. It is evident that the ANN model was unable to capture the “physics” of a heat exchanger or the feed heating process as a result of its poor performance in the out of normal scenarios. Further tuning by way of alternate activation functions and regularisation techniques had little effect on the ANN model performance. The ANN model was able to accurately predict an out of normal case only when it was trained to do so. This was achieved by augmenting the original training data with the inputs and results from the FLOWNEX SE® model for the same case. The conclusion drawn from this study is that this type of simple ANN model is able to predict plant performance so long as it is trained for it. The validity of the prediction is highly dependent on the integrity of the training data. Operating outside the range which the model was trained for will result in inaccurate predictions. It is recommended that out of normal scenarios commonly experienced by the plant be synthesised by engineering modelling tools like FLOWNEX® SE to augment the historic plant data. This provides a wider spectrum of training data enabling more generalised and accurate predictions from the ANN model.

Power grids have rapidly evolved into Smart grids and are heavily dependent on Supervisory Control and Data Acquisition (SCADA) systems for monitoring and control. However, this evolution increases the susceptibility of the remote (VMs, VPNs) and physical interfaces (sensors, PMUs LAN, WAN, sub-stations power lines, and smart meters) to sophisticated cyberattacks. The continuous supply of power is critical to power generation plants, power grids, industrial grids, and nuclear grids; the halt to global power could have a devastating effect on the economy's critical infrastructures and human life. Machine Learning and Deep Learning-based cyberattack detection modeling have yielded promising results when combined as a Hybrid with an Intrusion Detection System (IDS) or Host Intrusion Detection Systems (HIDs). This thesis proposes two cyberattack detection techniques; one that leverages Machine Learning algorithms and the other that leverages Artificial Neural networks algorithms to classify and detect the cyberattack data held in a foundational dataset crucial to network intrusion detection modeling. This thesis aimed to analyze and evaluate the performance of a Hybrid Machine Learning (ML) and a Hybrid Deep Learning (DL) during ingress packet filtering, class classification, and anomaly detection on a Smart grid network.

La recherche sur le photovoltaïque vise à réduire le prix par watt de puissance électrique générée. Des efforts considérables sont menés pour rechercher de nouveaux matériaux et des conceptions qui repoussent les limites des cellules solaires existantes. Le développement récent de matériaux et nanostructures complexes pour les cellules solaires nécessite des efforts plus importants pour mener à bien leur caractérisation et leur modélisation. Cette thèse porte sur la caractérisation optique, la modélisation et l'optimisation de la conception d'architectures de cellules solaires de pointe.Les mesures optiques sont utilisées pour la caractérisation rapide et non destructive des échantillons texturés pour les applications photovoltaïques. Les textures de surface améliorent le piégeage de la lumière et sont donc souhaitées pour améliorer les performances des cellules solaires. D'autre part, ces textures rendent la caractérisation optique plus difficile et des efforts plus importants sont nécessaires non seulement pour la mesure optique elle-même mais également pour la modélisation et l'interprétation ultérieure des données obtenues. Dans ce travail, nous démontrons que nous sommes en mesure d'utiliser des méthodes optiques pour étudier les textures pyramidales très répandues ainsi que les réseaux de nanofils de silicium à orientation aléatoire dont l'analyse est très difficile.Premièrement, nous nous sommes concentrés sur l'étude optique de diverses surfaces pyramidales et de leur impact sur les performances des cellules silicium à hétérojonction. Nous avons constaté que les angles au sommet des pyramides, préparées à l'aide de différentes conditions de texturation, diffèrent de la valeur théorique de 70.52° attendue pour le silicium cristallin. Cette modification de l'angle au sommet est expliquée par la présence, sur les facettes pyramidales, de terrasses monoatomiques régulières, observées par microscopie électronique à transmission de résolution atomique. L'impact d'une variation de l’angle au sommet sur les épaisseurs des couches minces déposées est étudié et les conséquences sur l'efficacité des cellules solaires résultantes sont discutées. Un modèle optique développé pour le calcul de la réflectance et de l'absorption des couches minces en multicouches sur surfaces pyramidales a permis l’optimisation de la conception de la cellule solaire pour un angle au sommet pyramidal donné.L'ellipsométrie matricielle Mueller a été utilisée in-situ pour caractéiser le processus de croissance - par méthode vapeur-liquide-solide activée par plasma - des nanofils de silicium. Nous avons développé un modèle optique facile à utiliser, qui, à notre connaissance, est le premier modèle utilisant des données ellipsométriques expérimentales pour contrôler le procédé de croissance, en phase vapeur-liquide-solide assisté par plasma, des nanofils. La dépendance linéaire observée du dépôt de matériau de silicium avec le temps de dépôt nous permet de suivre le processus de fabrication in situ et de contrôler la qualité du matériau. / Research in photovoltaics aims at lowering the price per watt of generated electrical power. Substantial efforts aim at searching for new materials and designs which can push the limits of existing solar cells. The recent development of complex materials and nanostructures for solar cells requires more effort to be put into their characterization and modeling. This thesis focuses on optical characterization, modeling, and design optimization of advanced solar cell architectures.Optical measurements are used for fast and non-destructive characterization of textured samples for photovoltaic applications. Surface textures enhance light-trapping and are thus desired to improve the solar cell performance. On the other hand, these textures make optical characterization more challenging and more effort is required for both, the optical measurement itself and subsequent modeling and interpretation of obtained data. In this work, we demonstrate that we are able to use optical methods to study the widely used pyramidal textures as well as very challenging randomly oriented silicon nanowire arrays.At first, we focused on the optical study of various pyramidal surfaces and their impact on the silicon heterojunction solar cell performance. We have found that vertex angles of pyramids prepared using various texturing conditions vary from the theoretical value of 70.52° expected from crystalline silicon. This change of the vertex angle is explained by regular monoatomic terraces, which are present on pyramid facets and are observed by atomic resolution transmission electron microscopy. The impact of a vertex angle variation on the thicknesses of deposited thin films is studied and the consequences for resulting solar cell efficiency are discussed. A developed optical model for calculation of the reflectance and absorptance of thin film multi-layers on pyramidal surfaces enabled a solar cell design optimization, with respect to a given pyramid vertex angle.In-situ Mueller matrix ellipsometry has been applied for monitoring the silicon nanowire growth process by plasma-enhanced vapor-liquid-solid method. We have developed an easy-to-use optical model, which is to our knowledge a first model fitting the experimental ellipsometric data for process control of plasma-assisted vapor-liquid-solid grown nanowires. The observed linear dependence of the silicon material deposition on the deposition time enables us to trace the fabrication process in-situ and to control material quality.

Mutations in the BRCA1 and BRCA2 genes are known risk factors and drivers of breast and ovarian cancers. So far, few studies have been focused on understanding the differences in transcriptome and functional landscapes associated with the disease (breast vs. ovarian cancers), gene (BRCA1 vs. BRCA2), and mutation type (germline vs. somatic). In this study, we were aimed at systemic evaluation of the association of BRCA1 and BRCA2 germline and somatic mutations with gene expression, disease clinical features, outcome, and treatment. We performed BRCA1/2 mutation centered RNA-seq data analysis of breast and ovarian cancers from the TCGA repository using transcriptome and phenotype 'portrayal' with multi-layer self-organizing maps and functional annotation. The results revealed considerable differences in BRCA1- and BRCA2-dependent transcriptome landscapes in the studied cancers. Furthermore, our data indicated that somatic and germline mutations for both genes are characterized by deregulation of different biological functions and differential associations with phenotype characteristics and poly(ADP-ribose) polymerase (PARP)-inhibitor gene signatures. Overall, this study demonstrates considerable variation in transcriptomic landscapes of breast and ovarian cancers associated with the affected gene (BRCA1 vs. BRCA2), as well as the mutation type (somatic vs. germline). These results warrant further investigations with larger groups of mutation carriers aimed at refining the understanding of molecular mechanisms of breast and ovarian cancers.

No description available.

Die Protonentherapie von Aderhautmelanomen wird vor allem für die Behandlung von Tumoren nahe kritischer Strukturen (Sehnerv) und bei großen Tumoren angewandt. Dabei ist die begrenzte Reichweite der Protonen vorteilhaft, die scharf begrenzte Dosisfelder im Auge ermöglicht, und das an den Tumor grenzende gesunde Gewebe bestmöglich schont. Daher erfolgt die Positionierung der Patienten und der Strahlenfelder in der Augentumortherapie, wie auch die regelmäßigen Konstanzprüfungen, mit einer Reichweitengenauigkeit in Wasser von 0,1 mm. Mit einem Multi-Leaf Faraday Cup (MLFC) kann die Reichweite der Protonen in kurzer Zeit sehr genau gemessen werden. Dabei misst der MLFC die differentielle Fluenz der Protonenstrahlen, also das Reichweitenprofil. Er besteht aus einem Stapel Folien, abwechselnd leitend und isolierend. Eindringende Protonen deponieren eine zusätzliche Ladung in der Folie in der sie stoppen. Durch eine gleichzeitige Strommessung an allen Folien misst der MLFC relativ schnell die Reichweite der Protonen. Aufgabe dieser Arbeit ist es, einen MLFC entsprechend den Anforderungen der Augentumortherapie zu entwickeln, aufzubauen und mögliche Anwendungspotentiale zu untersuchen. Dafür wurden Monte-Carlo-Rechnungen mit MCNPX 2.6 und SRIM durchgeführt, verschiedene Folienstapel an Luft und im Vakuum untersucht, verschiedene Messelektroniken zur gleichzeitigen Messung von Strömen im pA-Bereich in vielen Kanälen getestet, ein Absorbersystem für einen variablen Messbereich von 30 MeV bis 70 MeV aufgebaut und die entsprechende Mess- und Steuersoftware in LabVIEW 2011 entwickelt. Es wurde die Genauigkeit der Reichweitenmessungen untersucht und gezeigt, dass der MLFC durch seine Mobilität eine schnelle Energiebestimmung an unterschiedlichen Experimentierplätzen erlaubt. In der Therapie ist neben der einfachen Bestimmung der maximalen Reichweite der Protonen auch die regelmäßige Kontrolle der Modulation der ausgedehnten Bragg-Kurven möglich. / Proton therapy of uveal melanomas is primarily used for the treatment of tumors near critical structures (optic nerve) and in large tumors. The great advantage of protons is their sharply limited range in tissue, which leads to sharp defined dose fields in the eye and the dose absorbed by the healthy tissue around the tumor can be reduced. Therefore, the positioning of the patient and the radiation fields, as well as the regular control measurements in the eye tumor therapy requires an accuracy of 0.1 mm in water. A Multi-Leaf Faraday Cup (MLFC) gives the opportunity to measure the proton range relatively fast and accurate. The MLFC measures the differential fluence, which means the range profile of the proton beam. It consists of a stack of sheets, alternating conductive and insulating, and the penetrating protons bring their additional charge into the sheet in which they stop. By measuring the corresponding current in each conducting sheet at the same time, the MLFC can quickly measure the range of the protons. The task of this work is to develop a MLFC with respect to the requirements of the eye tumor therapy and to explore possible application potentials. Therefore, Monte Carlo calculations with MCNPX 2.6 and SRIM were conducted, various foil stacks were studied in air and in vacuum, different measurement electronics for measuring currents in the pA range in many channels simultaneously were tested, a system of degraders for a variable measuring range from 30 MeV to 70 MeV was developed and the corresponding measurement and control software was written in LabVIEW 2011. The accuracy of the range measurements was examined and it was shown that a quick energy measurement at different target stations can be made by the MLFC due to its mobility. In therapy, in addition to the determination of the maximum range of the proton beam, the regular monitoring of the modulation of the extended Bragg-curves is in principle possible.