Neste trabalho foi desenvolvido um novo instrumento para monitoramento computadorizado da curvatura da região central anterior da córnea humana durante cirurgias refrativas. Através da projeção de um anel iluminado na córnea, imagens dos reflexos são digitalizadas e processadas. Algoritmos baseados em técnicas de visão computacional e geometria óptica determinam a curvatura da região central (4mm), com alta precisão e desempenho. Mapas coloridos com códigos de cor em dioptrias (proporcionais ao inverso do raio da curvatura) são gerados para auxiliar o oftalmologista cirurgião no diagnóstico. / We have developed a new instrument for computer-aided monitoring of the central region of the anterior portion of the human cornea during refractive surgeries. By projecting a circular iluminated object on the cornea, reflected images are captured by a CCD, digitezed and processed. Algorithms based on computer vision technics and optical geometry compute the curvature of a 3 to 4 mm region, with righ precision and performance. Color-coded maps based on diopters (proportional to the inverse of the radius of curvature) are generated for ease of diagnose.

Usualmente a análise de sobrevivência considera a modelagem da função da taxa de falha ou função de risco. Uma alternativa a essa visão é estudar a vida residual, que em alguns casos é mais intuitiva do que a função de risco. A vida residual é o tempo de sobrevida adicional de um indivíduo que sobreviveu até um dado instante t0. Este trabalho descreve técnicas semiparamétricas e não paramétricas para estimar a média e a mediana de vida residual em uma população, testes para igualdade dessas medidas em duas populações e também modelos de regressão. Tais técnicas já foram testadas anteriormente em dados com baixa presença de censura; aqui elas são aplicadas a um conjunto de dados de pacientes com insuficiência cardíaca que possui uma alta quantidade de observações censuradas. / Usually, survival analysis is based on the modeling of the hazard function. One alternative approach is to consider the residual life, which would be more intuitive than the hazard function. Residual lifetime is the remaining survival time of a person given he or she survived a given time point t0. We describe semiparametric and non-parametric techniques for mean and median residual life estimation in a one-sample population, as well as tests for two-sample cases and regression models. Such techniques were previously tested for moderate censored data; here we apply them to heart-failure patients data with a high rate of censoring.

Na construção de cascos resistentes de submarinos são utilizados, frequentemente, os processos de conformação a frio e de soldagem. Estes processos produzem na estrutura deformações plásticas permanentes originando tensões residuais. A presença das tensões residuais é equivalente a introduzir uma pré-carga inicial na estrutura, o que acelera o processo de plastificação, reduzindo à capacidade de resistência da estrutura à pressão hidrostática. Para quantificar esta redução foi realizado, inicialmente, um estudo considerando a presença das tensões residuais devido à conformação a frio das chapas do casco e do flange das cavernas, para submarinos com 6, 8 e 10 m de diâmetro, em aço HY100. Para isso, um modelo não-linear foi produzido considerando não-linearidades geométricas e de material. Complementarmente, também foi estudada a influência de perfis de tensões residuais definidos a partir de resultados experimentais na redução da pressão de colapso do casco resistente do submarino espanhol S-80. Estes perfis consideram a presença simultânea de tensões residuais de conformação e de soldagem. Em todos os modelos estudados, as tensões residuais foram introduzidas no modelo numérico utilizando o comando INISTATE disponível no software comercial Ansys. Este comando é frequentemente utilizado na literatura em modelos numéricos envolvendo tensões residuais e foi validado utilizando três modelos de referência disponíveis na literatura. Ao final, pôde-se verificar que a presença das tensões residuais acelera a plastificação do casco resistente e reduz a pressão de colapso em até 5%, sendo a tensão residual de conformação a que mais contribuí nesta redução. De qualquer forma, pôde-se concluir que a influência das tensões residuais é pequena quando comparada com a pressão de colapso obtida para cada casco resistente analisado. / During the manufacturing of submarine pressure hull are often applied processes like cold forming and welding. Those processes lead to permanent plastic deformations which are associated with residual stresses. The presence of residual stresses is equivalent to the introduction of an initial pre-load in the structure, which accelerates the plastification process, decreasing hull pressure resistance. To quantify this reduction, a case study that considers residual stresses due to cold forming on hull and flange plates has been performed. The study encompasses hull diameters of 6, 8 and 10 m, made of HY100 steel. A nonlinear model has been done, considering material and geometric non-linearity. Complementarily, the influence of experimental residual stresses profiles on the reduction of collapse pressure of the Spanish S-80 submarine has been studied. These profiles consider the simultaneous presence of residual stresses due to cold forming and welding. In all studied models, the residual stresses have been introduced in the numerical models through INISTATE Ansys software command. This command has been validated using three reference models available in open source literature. In the end, it has been possible to verify that the presence of residual stresses increase the hull plastification and reduces the collapse pressure up to 5%, being the cold forming induced stress which most contributes to this reduction. Finally, it could be concluded, in the end of the study, that the influence of the residual stresses is small when compared with the collapse pressure obtained for the analyzed pressure hulls.

Na construção de cascos resistentes de submarinos são utilizados, frequentemente, os processos de conformação a frio e de soldagem. Estes processos produzem na estrutura deformações plásticas permanentes originando tensões residuais. A presença das tensões residuais é equivalente a introduzir uma pré-carga inicial na estrutura, o que acelera o processo de plastificação, reduzindo à capacidade de resistência da estrutura à pressão hidrostática. Para quantificar esta redução foi realizado, inicialmente, um estudo considerando a presença das tensões residuais devido à conformação a frio das chapas do casco e do flange das cavernas, para submarinos com 6, 8 e 10 m de diâmetro, em aço HY100. Para isso, um modelo não-linear foi produzido considerando não-linearidades geométricas e de material. Complementarmente, também foi estudada a influência de perfis de tensões residuais definidos a partir de resultados experimentais na redução da pressão de colapso do casco resistente do submarino espanhol S-80. Estes perfis consideram a presença simultânea de tensões residuais de conformação e de soldagem. Em todos os modelos estudados, as tensões residuais foram introduzidas no modelo numérico utilizando o comando INISTATE disponível no software comercial Ansys. Este comando é frequentemente utilizado na literatura em modelos numéricos envolvendo tensões residuais e foi validado utilizando três modelos de referência disponíveis na literatura. Ao final, pôde-se verificar que a presença das tensões residuais acelera a plastificação do casco resistente e reduz a pressão de colapso em até 5%, sendo a tensão residual de conformação a que mais contribuí nesta redução. De qualquer forma, pôde-se concluir que a influência das tensões residuais é pequena quando comparada com a pressão de colapso obtida para cada casco resistente analisado. / During the manufacturing of submarine pressure hull are often applied processes like cold forming and welding. Those processes lead to permanent plastic deformations which are associated with residual stresses. The presence of residual stresses is equivalent to the introduction of an initial pre-load in the structure, which accelerates the plastification process, decreasing hull pressure resistance. To quantify this reduction, a case study that considers residual stresses due to cold forming on hull and flange plates has been performed. The study encompasses hull diameters of 6, 8 and 10 m, made of HY100 steel. A nonlinear model has been done, considering material and geometric non-linearity. Complementarily, the influence of experimental residual stresses profiles on the reduction of collapse pressure of the Spanish S-80 submarine has been studied. These profiles consider the simultaneous presence of residual stresses due to cold forming and welding. In all studied models, the residual stresses have been introduced in the numerical models through INISTATE Ansys software command. This command has been validated using three reference models available in open source literature. In the end, it has been possible to verify that the presence of residual stresses increase the hull plastification and reduces the collapse pressure up to 5%, being the cold forming induced stress which most contributes to this reduction. Finally, it could be concluded, in the end of the study, that the influence of the residual stresses is small when compared with the collapse pressure obtained for the analyzed pressure hulls.

When an aluminum alloy plate after rolling, non-uniform residual stress distributions existed inside the plate and defects, such as edge wave, middle wave, of the plate will be induced. Usually, a levelling process will be adopted to modify the plate flatness. By numerically simulating the tension levelling process, the purpose of this thesis is to understand the final dimensions and the residual stress distribution of the aluminum plate subjected to the tension levelling process. This study used the finite element method as the basic theory of the numerical simulation. A 3-D model of a cold-rolled plate with a side wave, subjected to tension levelling process was constructed. Then, the effects of the variations of the tensile ratio and residual stress distribution after rolled on the residual stress distribution after levelling and the improvement of flatness were studied. The simulation results showed that in the wave region, the tension levelling process could eliminate more than 90% of the residual stress, in the flat region was up to 80%.Also, after leveling, the residual stress distribution in the flat region was more uniform than the wave region. After-rolled residual stresses at the wave region affected the final peak position of the wave and the stress eliminated ratio of the wave region, but showed no significant effect on the final plate width and the residual strains. After-rolled residual stresses at the flat region affected the stress elimination ratio of the flat region only. The tensile ratio would affect the plate flatness, the plate width, stress elimination ratio, and the maximum residual stress. The higher of the tensile ratio, the more flatness of the plate would be obtained, but the higher residual strain would be induced and caused the lesser range of available plate.

ABSTRACT AS-4 carbon fibers reinforced polyetheretherketone (PEEK) composite materials have been widely used in aerospace industry because of longer fatigue life, high specific stiffness and strength. The thesis is aimed to investigate the residual strength, residual stiffness and mechanical properties of thermoplastic AS-4/PEEK composite laminates subjected to tension-tension (T-T) cyclic loading at room temperature. We adopt modified diaphragm forming method by controlling temperature, pressure, vacuum and time conditions according to the obtained beast curing process to form composite laminates of low crystallinity, transcrystallinity and good fiber / matrix interfaces. Two common type of laminates are used, such as cross-ply [0/90]4S and quasi-isotropic [0/+45/90/-45]2S. Static tension test is performed to measure the elastic modulus and ultimate strength. And T-T fatigue test is conducted with maximum stress of 60% and 80% ultimate strength to find the residual strength and stiffness. Then, through the observation of failure surfaces of composite laminates we understand the failure initiation and mechanism by Scanning Electron Microscope (SEM). The results of experiment can be concluded as follows. The ultimate strength, elastic modulus and fatigue strength of cross-ply composite laminates are larger than those of quasi-isotropic. As centrally notched, the net area of the specimen is reduced, the ultimate strength and fatigue strength of composite materials are lower. The residual strength, adopted to describe the damage process, is monotonically decreasing with increasing of applied cycles. It is found that the residual strength of cross-ply laminates is larger than that of quasi-isotropic laminates. However, the residual stiffness has little change with increasing of applied cycles.

This bachelor thesis has been carried out at GKN Aerospace. GKN is a member of European Space Agency, designing and manufacturing rocket-nozzles for the Ariane rockets. The manufacturing process entails many welds. Weld-simulations have been made to investigate stresses and plastic strains on simplified geometries. Plastic strains have been evaluated parallel and normal to the weld for plate geometries of shell-elements with rectangular cross-section and sandwich-cross-section, using the FEM-program MSC.marc. Results shows that plate width and length have negligible effect on the plastic strains when one weld is made. A comparison between a sandwich-sector cone and a sandwich plate was made, to investigate how plastic strains and stresses were affected of geometry. Plastic strains and stresses parallel the weld are the same. Plastic strains and stresses normal the weld are affected by changing geometry. Studies on differences in stresses between solid and shell elements propose use of solid elements near the weld region, if stresses are of interest.

Neste trabalho foi desenvolvido um novo instrumento para monitoramento computadorizado da curvatura da região central anterior da córnea humana durante cirurgias refrativas. Através da projeção de um anel iluminado na córnea, imagens dos reflexos são digitalizadas e processadas. Algoritmos baseados em técnicas de visão computacional e geometria óptica determinam a curvatura da região central (4mm), com alta precisão e desempenho. Mapas coloridos com códigos de cor em dioptrias (proporcionais ao inverso do raio da curvatura) são gerados para auxiliar o oftalmologista cirurgião no diagnóstico. / We have developed a new instrument for computer-aided monitoring of the central region of the anterior portion of the human cornea during refractive surgeries. By projecting a circular iluminated object on the cornea, reflected images are captured by a CCD, digitezed and processed. Algorithms based on computer vision technics and optical geometry compute the curvature of a 3 to 4 mm region, with righ precision and performance. Color-coded maps based on diopters (proportional to the inverse of the radius of curvature) are generated for ease of diagnose.

Usualmente a análise de sobrevivência considera a modelagem da função da taxa de falha ou função de risco. Uma alternativa a essa visão é estudar a vida residual, que em alguns casos é mais intuitiva do que a função de risco. A vida residual é o tempo de sobrevida adicional de um indivíduo que sobreviveu até um dado instante t0. Este trabalho descreve técnicas semiparamétricas e não paramétricas para estimar a média e a mediana de vida residual em uma população, testes para igualdade dessas medidas em duas populações e também modelos de regressão. Tais técnicas já foram testadas anteriormente em dados com baixa presença de censura; aqui elas são aplicadas a um conjunto de dados de pacientes com insuficiência cardíaca que possui uma alta quantidade de observações censuradas. / Usually, survival analysis is based on the modeling of the hazard function. One alternative approach is to consider the residual life, which would be more intuitive than the hazard function. Residual lifetime is the remaining survival time of a person given he or she survived a given time point t0. We describe semiparametric and non-parametric techniques for mean and median residual life estimation in a one-sample population, as well as tests for two-sample cases and regression models. Such techniques were previously tested for moderate censored data; here we apply them to heart-failure patients data with a high rate of censoring.

Indiana University-Purdue University Indianapolis (IUPUI) / Machine learning models are rising every day. Most of the Computer Vision oriented machine learning models arise from Convolutional Neural Network’s(CNN) basic structure. Machine learning developers use CNNs extensively in Image classification, Object Recognition, and Image segmentation. Although CNN produces highly compatible models with superior accuracy, they have their disadvantages. Estimating pose and transformation for computer vision applications is a difficult task for CNN. The CNN’s functions are capable of learning only shift-invariant features of an image. These limitations give machine learning developers motivation towards generating more complex algorithms. Search for new machine learning models led to Capsule Networks. This Capsule Network was able to estimate objects’ pose in an image and recognize transformations to these objects. Handwritten digit classification is the task for which capsule networks are to solve at the initial stages. Capsule Networks outperforms all models for the MNIST dataset for handwritten digits, but to use Capsule networks for image classification is not a straightforward multiplication of parameters. By replacing the Capsule Network’s initial layer, a simple Convolutional Layer, with complex architectures in CNNs, authors of Residual Capsule Network achieved a tremendous change in capsule network applications without a high number of parameters. This thesis focuses on improving this recent Residual Capsule Network (RCN) to an extent where accuracy and model size is optimal for the Image classification task with a benchmark of the CIFAR-10 dataset. Our search for an exemplary capsule network led to the invention of RCN2: Residual Capsule Network 2 and RCNX: Residual Capsule NeXt. RCNX, as the next generation of RCN. They outperform existing architectures in the domain of Capsule networks, focusing on image classification such as 3-level RCN, DCNet, DC Net++, Capsule Network, and even outperforms compact CNNs like MobileNet V3. RCN2 achieved an accuracy of 85.12% with 1.95 Million parameters, and RCNX achieved 89.31% accuracy with 1.58 Million parameters on the CIFAR-10 benchmark.