Made available in DSpace on 2018-08-02T00:00:39Z (GMT). No. of bitstreams: 1 tese_10533_Versao_Mestrado_Leonardo.pdf: 28216764 bytes, checksum: 15d3f7dcdd4407dda2903fdb56efe45e (MD5) Previous issue date: 2016-12-16 / Com o aumento da capacidade de processamento dos dispositivos, imagens podem ser usadas para analisar cenas e extrair informações tridimensionais dos pixels em um espaço de tempo bem menor do que se conseguia fazer há alguns anos atrás. Esse processo de recuperação da informação tridimensional do meio recebe o nome de reconstrução 3D. Estimar com precisão a profundidade de regiões homogêneas de uma imagem ainda é um desafio na área. Neste trabalho, foram propostas duas metodologias de reconstrução 3D densa aproximando-se regiões homogêneas por planos e por uma superfície obtida a partir da triangulação de Delaunay. Essa abordagem tem como objetivo apresentar um bom compromisso entre precisão e velocidade. O trabalho está focado apenas na etapa de reconstrução, sendo a estimativa de parâmetros intrínsecos considerada obtida por calibração e os parâmetros extrínsecos resultantes de um processo de tracking da câmera. Para a estimação da reconstrução, é utilizado um conjunto de 10 imagens que possuem certa sobreposição com uma imagem de referência. O correspondente de cada pixel da imagem de referência é procurado nas 10 imagens resultando em uma estimativa esparsa após uma etapa de fusão de informações e filtragem de possíveis outliers. É possível diminuir o tempo de processamento do algoritmo, utilizando-se uma pirâmide gaussiana de multiresolução. As regiões homogêneas são identificadas com a técnica SRM, tendo sua profundidade estimada a partir da nuvem de pontos esparsa reconstruída a priori utilizando uma das duas metodologias propostas neste trabalho: aproximação por planos ou por uma superfície obtida a partir da triangulação de Delaunay. Os resultados obtidos foram satisfatórios segundo os critérios de precisão, recall e tempo de processamento. Comparando-se com o DTAM, uma técnica de tempo real, os resultados obtiveram melhor precisão, principalmente em regiões homogêneas. O maior tempo de processamento é devido ao hardware de menor capacidade quando comparado ao caso do DTAM. Além disso, o código implementado não está otimizado. Desta forma, é possível se obter velocidades maiores de reconstrução com o método proposto. De acordo com as avaliações feitas neste trabalho, os resultados obtidos podem ser considerados promissores.

Afin de mieux comprendre les mécanismes du trafic cellulaire, les processus de transport des substrats vacuolaires à travers le tonoplaste, le stockage des métabolites et leur dégradation, une analyse globale et exhaustive du protéome vacuolaire d'Arabidopsis thaliana a été réalisée. La connaissance de la localisation subcellulaire des protéines permet de mieux comprendre la fonction des organelles et la compartimentation du métabolisme des plantes. Mais la description précise du protéome d'un organite nécessite d'identifier clairement les véritables protéines résidantes du compartiment étudié. Une tâche si précise est complexe puisqu'elle nécessite la mise en place d'une préparation d'organites purs et homogènes. Pour y parvenir, un protocole de purification de vacuoles à partir de protoplastes isolés de cellules en culture sur un gradient de densité de Ficoll a été amélioré. La combinaison de plusieurs approches de protéomique a permis d'identifier les protéines présentes dans les fractions vacuolaires soluble et membranaire de façon quantitative et fonctionnelle. Les différentes approches ont ainsi mis en évidence des associations et mécanismes moléculaires complexes qui régissent les différentes activités vacuolaires. Cette protéothèque de référence constitue une base pour étudier la dynamique du protéome vacuolaire en réponse à plusieurs stress incluant les métaux lourds. Plusieurs méthodes sans a priori et ciblée ont été proposé afin d'étudier l'impact du cadmium sur la vacuole, ce compartiment cellulaire clé de la détoxication. / To better understand the mechanisms governing cellular traffic, transport process of substrates across the tonoplast, storage of various metabolites and their ultimate degradation, a comprehensive and thorough analysis of Arabidopsis thaliana vacuolar proteome was performed. Protein subcellular localization knowledge is an important step toward assigning functions of organelles and plant metabolism compartmentation. But confident description of proteome organelle content requires clear identification of the true resident proteins of the studied compartment. This task involves pitfalls and requires that either organelle preparations are free of contaminants or that techniques are used to discriminate between genuine organelle residents and contaminating proteins. To achieve this, vacuoles purification protocol from protoplasts on a Ficoll density gradient has been improved. The combination of several proteomic approaches attempt to present soluble and membrane vacuolar proteins in a quantitative and functional manner. Different approaches have thus shown associations and complex molecular mechanisms that govern the various vacuolar activities. The constitute proteins library provides references to study the vacuolar proteome dynamics in response to different stresses including heavy metals. Many methods without a priori or targeted were proposed to study the impact of cadmium on the vacuole, the key cell compartment of detoxification. Proteomics provides powerful tools for characterizing the protein contents of vacuoles during cadmium stress.

A learning task with thousands of training examples in Support Vector Machine (SVM) demands large amounts of memory and time requirements. SVMlight by Dr. Thorsten Joachims has been implemented in C using a fast optimizing algorithm for handling thousands of such support vectors. SVMlight solves the problem of classification, pattern recognition, regression and learning ranking function. The C code also provides methods for XiAlpha estimation of error rate and precision. Implementing these two methods leads to generalized performance of Support Vector Machine even for computation intensive text classification functions. SVMlight code allows users to define their own kernel functions. The SVMlight software employs an efficient algorithm and minimizes the cost, but it still takes considerable amount of time for computing thousands of support vectors and training examples. This time can be still reduced by parallelizing the code. In our work we refined the SVMlight code by removing unnecessary iterations and rewriting it as cost efficient. Then we parallelized the code individually using two different types, OpenMP and POSIX Threads shared memory parallelism. The code is parallelized for these two methods on Intel’s C compiler for Linux 7.1 using hyper threading technology. The parallelized code is tested for protein structure prediction. Different types of Protein Sequences are tested on these methods by varying the number of training examples and support vectors. The time consumption and speedup are calculated for both OpenMP and Pthreads. Implementation of OpenMP and Pthreads together showed good increase in speedup.

Clopidogrel is an antiplatelet substance that prevents blood coagulation in the arteries. It is an inactive pro drug that becomes activated after first-pass metabolism by the liver. The active metabolite of clopidogrel is 2-oxoclopidogrel, which is unstable therefore pharmacokinetic data is obtained by measuring the inactive metabolite clopidogrel acid in plasma. Clopidogrel is taken orally in tablet form. The aim of this project was to develop a LC-MS/MS method for quantification of clopidogrel and its metabolite in plasma.   The method has been developed by optimizing the sample preparation. Different extraction procedures and extraction columns were tested, for example, by changing the extraction column from a C8 silica sorbent to Oasis HLB (a polymer sorbent). Different internal standards were evaluated as a result of discovering the signal suppression of the previous internal standard clopidogrel acid.  Flupentixol was found to be the best candidate.

Rotor Position information is essential in the operation of the Switched Reluctance Motor (SRM) for properly controlling its phase currents. This thesis uses Field Programmable Gate Array (FPGA) technology to implement a method to estimate the SRMs rotor position using the inverse inductance value of the SRMs phases. The estimated rotor position is given as input to the Commutator circuit, also implemented in the FPGA, to determine when torque-producing currents should be input in the SRM phase windings. The Estimator and Commutator design is coded using Verilog HDL and is simulated using Xilinx tools. This circuit is implemented on a Xilinx Virtex XCV800 FPGA system. The experimentally generated output is validated by comparing it with simulation results from a Simulink model of the Estimator. The performance of the FPGA based SRM rotor position estimator in terms of calculation time is compared to a digital signal processor (DSP) implementation of the same position estimator algorithm. It is found that the FPGA rotor position Estimator with a 5MHz clock can update its rotor position estimate every 7s compared to an update time of 50s for a TMS320C6701-150 DSP implementation using a commercial DSP board. This is a greater than 7 to one reduction in the update time.

This thesis presents the development of an in-wheel drive system consisting of an axial-flux switched-reluctance motor and a hub suspension. The motor is designed using Maxwell's stress tensor and numerical analysis techniques, including FEA and transient numerical simulations. A new integral inductance function is introduced that improves the accuracy of the motor model, and a new in-phase current-shaping technique is implemented using a fuzzy controller to extend the constant-power region of the motor. The hub suspension system is simulated using a half-car model with 6 degrees of freedom, and the overall torque, power, and efficiency of the drive system is calculated. A peak torque of 500[Nm] is developed at the high end of the drive system's speed range, and the hub suspension system is shown to eliminate the impact of the motor's increased unsprung mass on vehicle handling.

With the increasing environmental impact from the automotive industry, electric vehicles become more and more popular. This combined with the great breakthroughs in fast electronics the switched reluctance motor (SRM) has again gained popularity in recent years. Due to its cheap and rugged construction it is a good alternative to the permanent magnet motors and to the induction motor. The ́two main problems holding the SRM back are torque ripple and the acoustic noise generated from it. A lot of research is currently being performed in order to find a solution to these issues. This thesis has investigated different materials for the rotor in a high speed SRM. Different materials have been evaluated based on both mechanical and magnetic properties. This is done through simulations of the forces acting on the rotor combined with simulations of the magnetic field. The forces are simulate in the DASSULT SYSTEMS ABAQUS program and the magnetic field is simulate using AVL FIRE. Three different kinds of alloys are investigated, two different cobalt alloys are simulated as well as a silicon alloy with pure iron as a reference. The results show that the material needs to have a yield strength of at least 349 MPa to withstand the forces affecting the rotor. And that by using the high purity cobalt-iron alloy the generated torque could be increased with up to 20.9%, but with a cost increase of 3151.9% compared to the silicon alloy.

This thesis characterizes the transient thermal response of a 12/16 External Rotor Switched Reluctance Machine (ERSRM) for an E-bike application. A method for calculating coil capacitance based on machine design parameters was introduced and implemented into a standard commercial Lumped Parameter Thermal Network (LPTN). A sizing criterion was proposed for the cuboid number in a physically accurate LPTN coil model design. This sizing criterion considers the change in model size with motor speed or forced convection. The LPTN with a more accurate calculation of capacitance within the coil and a known number of cuboids in the coil was validated with experimental results. An analytical proof was provided that a small number of capacitances is not sufficient to model a typical power-dense coil design. The validated model was used to study the impact of a more accurate capacitance calculation method on motor temperature. Both overload and rated operation were investigated. During overload conditions, it was found that the standard capacitance calculation from commercial software massively underestimated the heating rate and peak temperature of the coil hot spot, even with the same number of cuboids. The capacitance of the rest of the motor was able to be varied and investigated for its effects on cooldown dynamics. It was found that for short-time transients the coil could be assumed to act adiabatically in this operating range. Operating points across the operating envelope for the motor under study were mapped to determine the region where the adiabatic assumption could be made. It was shown that a transition occurred where the adiabatic assumption ceases to be valid. / Thesis / Doctor of Philosophy (PhD) / This thesis deals with the thermal modelling of electric machines for traction applications using lumped parameter thermal modelling. A novel approach is presented for calculating and distributing thermal capacitance in motor coils. A 12/16 External Rotor Switched Reluctance Motor is characterized based on its transient thermal response and the novel methods proposed are validated. The sizing of a coil-based thermal model is discussed and a criterion for physical validity proposed. The validated model is used in a sensitivity analysis of coil and motor capacitances. For severe overload conditions and short periods, a result is obtained showing the coil can be modelled as adiabatic. Finally, a rated load condition is tested, and a transition is suggested between overload conditions and non-overload conditions.

Switched reluctance motors (SRMs) possess many desirable qualities for the long-term sustainability of electrified transportation such as cheap production costs and simple, robust configurations. However, high acoustic noise and torque ripple are two performance imperfections that have prevented the widespread implementation of SRMs. This thesis investigates design techniques to reduce the acoustic noise produced by an 8/6 SRM while also analyzing the impact each design has on the motor’s performance. The fundamentals of SRMs are discussed including the operating principles, modelling and control strategies. The multiphysics finite element analysis (FEA) toolchain used to accurately model acoustic noise and vibrations of SRMs is described. Using the network of FEA tools, nodal forces and natural frequencies of a four phase 8/6 SRM are analyzed to study the acoustic noise and vibration behaviours. The FEA process is validated experimentally by matching measured vibration modes and acoustic noise sound pressure level (SPL) with FEA numerical results. Through inspiration from an extensive literature review, various design techniques are applied to a baseline four phase 8/6 SRM and compared for both acoustic noise reduction and EM performance criteria. The investigated designs were split into two categories, stator-housing modifications that aim to increase the stiffness of the assembly and rotor modifications that aim to reduce the magnitude of radial forces while preserving performance. The best design strategies as determined by the comparative analysis were then further optimized to combine the best techniques together for the 8/6 SRM. The proposed structural improvements included the modifications of the stator yoke shape along with increasing the number of fastening components involved in the assembly. Additionally, an iterative procedure for the parametric modelling of windows introduced to the rotor poles is outlined. The best design considerations are combined to create the design of a novel 8/6 SRM which significantly reduces the acoustic noise produced by the motor with little impact to performance. / Thesis / Master of Applied Science (MASc)

Switched Reluctance Machines (SRMs) are gaining more attention due to their simple and rugged construction, low manufacturing cost, and high-speed operation capability. An electromagnetic model of the machine is needed in the design and analysis processes. The required accuracy level of the model depends mainly on the application. Designing an SRM is an iterative process. Usually, finite element method (FEM) is employed in all design stages, which might require extensive computation burden. The magnetic equivalent circuit (MEC) method is an alternative for typical FEM. MEC models require less computational resources and they can help determine the electromagnetic performance with a reasonable accuracy. The conventional MEC method can be challenging when modifying the motor geometry while conducting dynamic analysis with current control. This thesis proposes a reluctance mesh-based MEC model for SRMs that can overcome those challenges. Reluctance mesh-based MEC models are developed for 3-phase 6/4, 6/16, 12/8 SRMs and 4-phase 8/6, 8/10, and 16/12 SRMs. The implemented MEC-based modeling method is validated using FEM and experimental results. Acoustic noise and vibration is one of the shortcomings of an SRM. The radial force density in the airgap should be calculated before analyzing and mitigating acoustic noise and vibration. This thesis proposes a radial force density calculation method for SRMs using the proposed MEC model. Fourier series is used to calculate the harmonics of the radial force density. The results obtained from the MEC model are verified using FEM models. SRM is a promising candidate for electric propulsion systems. In the design process of an SRM, the motor geometry needs to be determined. This thesis applies the proposed MEC technique to the design process of a 3-phase 12/16 SRM for a high lift motor in the NASA Maxwell X-57 electric aircraft. The design is verified using the results computed from FEM. / Dissertation / Doctor of Philosophy (PhD) / Electric motors are utilized in our daily life in various applications such as washing machines, refrigerators, air conditioners, fans, vacuum cleaners, blenders, and many other devices and tools. Motors are widely used in residential, industrial, commercial, and transportation applications. Due to the environmental impact of burning fossil fuels, transportation systems are moving into electrified propulsion. Electric motors with lower cost and higher efficiency are on the path to replacing the traditional combustion engines in vehicles. Among the different electric motors available, switched reluctance motor (SRM) is becoming a promising candidate in future electrified transportation systems due to their simple construction. Developing a motor is a time-consuming and costly task. Therefore, it is essential to determine the characteristics of an SRM before manufacturing it. A mathematical framework is proposed in this thesis to address this problem. The proposed framework is capable of determining the characteristics of an SRM accurately.