Orientador: Pablo Siqueira Meirelles / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenharia Mecânica / Made available in DSpace on 2018-08-20T08:25:52Z (GMT). No. of bitstreams: 1 Piveta_Alessandro_M.pdf: 20999298 bytes, checksum: 60bf64ff246a3515b0455a01ccb8e817 (MD5) Previous issue date: 2012 / Resumo: No desenvolvimento de projetos mecânicos é comum o uso de valores determinísticos para definir parâmetros que descrevem o sistema, atribuindo elevados coeficientes de segurança a fim de contornar o problema de incertezas relacionadas aos valores das cargas externas aplicadas, da variabilidade das propriedades dos materiais, das condições ambientais, entre outros fatores que possam afetar o desempenho do sistema. Atualmente, com o desenvolvimento de alguns métodos, pode-se introduzir modelos estocásticos que consideram estas incertezas e variações, deixando assim o modelo mais próximo da realidade. Este trabalho analisa o efeito da variabilidade simultânea de determinados parâmetros no desempenho de dois tipos de suspensão: convencional e hidropneumática. Utilizou-se um modelo matemático que representa um veículo completo com sete graus de liberdade implementado no software Matlab®. É desenvolvida a formulação para modelar a suspensão hidropneumática considerando que o gás contido no acumulador sofre uma mudança de estado politrópico. Inicialmente são apresentadas as respostas dos modelos determinísticos devido a excitações transientes representadas por funções que descrevem uma lombada e um degrau com diferentes dimensões. Em seguida, é realizada a análise de sensibilidade da rigidez hidropneumática com o intuito de descobrir quais parâmetros possuem maior influência sobre este sistema. O Método de Monte Carlo é utilizado como solver estocástico nas simulações, tendo como objetivo avaliar a influência das variações dos parâmetros sobre a dinâmica vertical do veículo. A distribuição de probabilidade escolhida para cada parâmetro é a função gama. Por último, são analisadas as respostas resultantes dos modelos probabilísticos. As saídas avaliadas são: aceleração vertical do centro de gravidade, razão de amortecimento, valor da força de contato entre pneu e via e ângulo de rolagem da massa suspensa / Abstract: Deterministic values are largely used in mechanical design to define the parameters which describe the system. Large coefficients of safety are used because of the uncertainties related to these parameters. There are uncertainties related to the external loadings, to the variability of the material properties and to the environmental conditions, among other important parameters which affect the performance of the system. In the last decade, stochastic models have been used to consider these uncertainties and variations. The present work analyzes the effect of the simultaneous variability of certain parameters on the performance of the conventional and the hydropneumatic suspension of automotive vehicles. A mathematical model of a full vehicle with seven degrees of freedom has been used. This model was implemented in the software Matlab®. In the present formulation of the hydropneumatic suspension, it is considered that the gas contained in the accumulator undergoes a polytrophic process. First, the deterministic model of the vehicle is subjected to two transient excitations, namely a function describing a bump and a function describing a step with different dimensions. Then, an analysis of sensibility of the stiffness of the hydropneumatic suspension is performed, which aims to identify which parameters are the most influential on that system. Monte Carlo method is used as the stochastic solver throughout the simulations which study the influence of the parameter variability on the vertical dynamics of the vehicle. The Gamma function is used as the probability density function for each parameter. Finally, the response of the probabilistic models is studied. This work investigates the acceleration of the center of gravity, damping factor, pitch angle of the suspended mass and contact force between the tires and the road / Mestrado / Mecanica dos Sólidos e Projeto Mecanico / Mestre em Engenharia Mecânica

The Bayesian approach to logistic regression modelling for credit scoring is useful when there are data quantity issues. Data quantity issues might occur when a bank is opening in a new location or there is change in the scoring procedure. Making use of prior information (available from the coefficients estimated on other data sets, or expert knowledge about the coefficients) a Bayesian approach is proposed to improve the credit scoring models. To achieve this, a data set is split into two sets, “old” data and “new” data. Priors are obtained from a model fitted on the “old” data. This model is assumed to be a scoring model used by a financial institution in the current location. The financial institution is then assumed to expand into a new economic location where there is limited data. The priors from the model on the “old” data are then combined in a Bayesian model with the “new” data to obtain a model which represents all the available information. The predictive performance of this Bayesian model is compared to a model which does not make use of any prior information. It is found that the use of relevant prior information improves the predictive performance when the size of the “new” data is small. As the size of the “new” data increases, the importance of including prior information decreases

Over the last few decades, radiotherapy using high-energy proton beams over the range from 50 MeV to 250 MeV has been increasingly used and developed. Indeed, it offers the possibility to focus the dose in a very narrow area around the tumor cells. The tumor control is improved compared to radiotherapy using photon beams and the healthy cells around the tumor are not irradiated since the range of charged particles is limited. However, due to nuclear reactions of the incident charged particles in the tissue, secondary high-energy radiations, essentially photons and neutrons, are produced and irradiate the treatment room.<p>As a consequence, thick concrete shielding walls are placed around the treatment room to ensure that other people and workers received a dose as small as possible. The dose measurement is performed with specific dosemeters such as the WENDI-II, which gives a conservative estimation of the ambient dose equivalent up to 5 GeV. The dose in working areas may also be estimated by means of numerical calculations by using simulation codes of particle transport such as the GEANT4, MCNPX, FLUKA and PHITS Monte Carlo codes.<p>Secondary particle yields calculated with Monte Carlo codes show discrepancies when different physical models are used but are globally in good agreement with experimental data from the literature. Neutron and photon doses decrease exponentially through concrete shielding wall but the neutron dose is definitely the main component behind a wall with sufficient thickness. Shielding parameters, e.g. attenuation coefficients, vary as functions of emission angle (regarding the incident beam direction), incident proton energy, and target material and composition.<p>The WENDI-II response functions computed by using different hadronic models show also some discrepancies. Thermal treatment of hydrogen in the polyethylene composing the detector is also of great importance to calculate the correct response function and the detector sensitivity.<p>Secondary particle sources in a proton therapy facility are essentially due to losses in cyclotron and beam interactions inside the energy selection system, with the treatment nozzle components and the target - patient or phantom. Numerical and experimental results of the dose in mazes show a good agreement for the most of detection points while they show large discrepancies in control rooms. Indeed, statistical consistency is reached with difficulty for both experimental and calculated results in control rooms since concrete walls are very thick in this case.<p>/<p>La radiothérapie utilisant des faisceaux de protons d’énergie entre 50 MeV et 250 MeV s’est largement développée ces dernières années. Elle a l’immense avantage de pouvoir concentrer la dose due au faisceau incident de manière très efficace et très précise sur la tumeur, en épargnant les éventuels organes sains et sensibles aux radiations situés aux alentours. Cependant, des rayonnements « secondaires » très énergétiques sont créés par les réactions nucléaires subies par les protons lors de leur parcours dans les tissus, et peuvent sortir du patient. Des blindages entourant la salle de traitement et suffisamment épais doivent être présents afin que la dose reçue par les personnes se trouvant aux alentours soit la plus faible possible. La mesure de la dose se fait avec des dosimètres spécifiques et sensibles aux rayonnements de haute énergie, tels que le WENDI-II pour les neutrons. L’estimation de cette dose, et donc la modélisation des blindages, se fait également avec des codes de simulation numérique de transport de particules par les méthodes de Monte Carlo, tels que GEANT4, MCNPX, FLUKA et PHITS.<p>La production de rayonnements secondaires calculée à l’aide de codes Monte Carlo montre des écarts significatifs lorsque différents modèles d’interactions physiques sont utilisés, mais est en bon accord avec des données expérimentales de référence. L’atténuation de la dose due aux neutrons et aux photons secondaires à travers un blindage composé de béton est exponentielle. De plus, la dose due aux neutrons est clairement la composante dominante au-delà d’une certaine épaisseur. Les paramètres d’atténuation, comme par exemple le coefficient d’atténuation, dépendent de l’angle d’émission (par rapport à la direction du faisceau incident), de l’énergie des protons incidents et de la nature et la composition de la cible.<p>La fonction de réponse du dosimètre WENDI-II montre également des variations lorsque différents modèles physiques sont considérés dans les codes Monte Carlo. La prise en compte d’effets fins comme les états de vibration et de rotation des atomes d’hydrogène au sein du polyéthylène composant le détecteur se révèle essentielle afin de caractériser correctement la réponse du détecteur ainsi que sa sensibilité.<p>L’émission secondaire dans un centre de protonthérapie est essentiellement due aux pertes dans le cyclotron et aux interactions du faisceau avec les systèmes de sélection de l’énergie, les composants de la tête de tir et le patient (ou le fantôme). L’évaluation numérique de la dose dans les labyrinthes des différentes salles du centre montre un bon accord avec les données expérimentales. Tandis que pour les points de mesure dans leur salle de contrôle respective, de larges différences peuvent apparaitre. Ceci est en partie dû à la difficulté d’obtenir des résultats statistiquement recevables du point de vue expérimental, mais aussi numérique, au vu de l’épaisseur des blindages entourant les salles de contrôle. / Doctorat en Sciences de l'ingénieur / info:eu-repo/semantics/nonPublished

Title: Tests of independence for multivariate data Author: Bc. Michal Kudlík Department: Department of Probability and Mathematical Statistics Supervisor: Ing. Marek Omelka, PhD., Department of Probability and Mathema- tical Statistics Abstract: This thesis is an overview of tests of independence for multidimensi- onal data. The report includes tests on independence of categorical and conti- nuous random variables, tests assuming normal distribution of data, asymptotic nonparametric tests and permutation tests with application of the Monte Carlo method. This thesis shows the suitability of tests with properly chosen real data and checks significance level and compares the strength of the selected tests by simulation study while using appropriate statistical software. Based on the simu- lation study the thesis discusses an appropriateness of the use of different tests for different situations. Keywords: independence, permutation and asymptotic tests of independence, Monte Carlo method, simulation study 1

M.Sc. (Mathematical Statistics) / Comparison of two distributions via use of the quantile comparison function is carried out specifically from possibly censored data. A semi-parametric method which assumes linearity of the quantile comparison function is examined thoroughly for non-censored data and then extended to incorporate censored data. A fully nonparametric method to construct confidence bands for the quantile comparison function is set out. The performance of all methods examined is tested using Monte Carlo Simulation.

M.Sc. (Mathematical Statistics) / Stochastic Differential Equations (SDE’s) are commonly found in most of the modern finance used today. In this dissertation we use SDE’s to model a random phenomenon known as the short-term interest rate where the explanatory power of a particular short-term interest rate model is largely dependent on the description of the SDE to the real data. The challenge we face is that in most cases the transition density functions of these models are unknown and therefore, we need to find reliable and accurate alternative estimation techniques. In this dissertation, we discuss estimating techniques for discretely sampled continuous diffusion processes that do not require the true transition density function to be known. Moreover, the reader is introduced to the following techniques: (i) continuous time maximum likelihood estimation; (ii) discrete time maximum likelihood estimation; and (iii) estimating functions. We show through a Monte Carlo simulation study that the parameter estimates obtained from these techniques provide a good approximation to the estimates obtained from the true transition density. We also show that the bias in the mean reversion parameter can be reduced by implementing the jackknife bias reduction technique. Furthermore, the data analysis carried out on South-African interest rate data indicate strongly that single factor models do not explain the variability in the short-term interest rate. This may indicate the possibility of distinct jumps in the South-African interest rate market. Therefore, we leave the reader with the notion of incorporating jumps into a SDE framework.

La filière solaire thermodynamique, et plus précisément la technologie des centrales à tour, est l'une des ressources d'énergies renouvelables les plus prometteuses. Le travail de thèse s’inscrit dans le cadre général de l’augmentation des rendements de centrales solaires à tour où l’augmentation de l'efficacité et de la température du récepteur joue un rôle clé. L’originalité de la thèse réside dans l’étude du rôle des sélectivités spectrale et directionnelle sur le rendement de récepteurs surfaciques à haute température. L’objectif poursuivit a été d’améliorer simultanément la géométrie, le matériau, la rugosité et la microstructuration des récepteurs surfaciques pour créer une sélectivité permettant l’augmentation des rendements. La première étape a consisté à l’élaboration d’un modèle approché d’une centrale solaire pour quantifier l’influence du rendement du récepteur sur le rendement de la centrale. Dans un second temps, un modèle de récepteur détaillé incluant la résolution précise des transferts radiatifs (grâce à la méthode Monte Carlo) a été développé pour déterminer le gain en rendement que l’on peut obtenir grâce à la sélectivité spectrale et directionnelle. Ensuite, la modification des propriétés radiatives des surfaces par des microstructures unipériodiques et bipériodiques a été étudiée dans l'objectif d'augmenter la sélectivité spectrale du récepteur. Cette étape a nécessité la résolution des équations de Maxwell par la méthode RCWA. La microstructure retenue et qui offre le meilleur potentiel possède un relief pyramidal bipériodique. L’optimisation de la période et de la hauteur de ce relief permet d’obtenir une absorptivité solaire maximale égale à 0,98. / Concentrated solar power (CSP) technology is an alternative for renewable thermal energy generation and a promising source of energy. Managing the optical properties of a cavity solar receiver to create spectral and directional selectivities is a solution to improve receiver and solar power plant efficiencies in order to reduce cost. The two main steps of the development are to quantify the potential gain that may result from theoretical selectivities and then the optimization of microstructuration to increase as possible the spectral selectivity. At first, a calculation code was built in order to quantify the influence of receiver on the global solar power plant efficiency. Then a parametric study which takes into account thegeometry, the material, the roughness and the microstructure of the receiver was conducted. At last, an optimization of microstructuration with uniperiodic and biperiodic structures was studied. As a conclusion the best structure is a biperiodic pyramidal relief which offers a huge gain of absorptivity.

La fabrication des structures composites conduit à une variabilité élevée de ses paramètres mécaniques. La thèse a comme objectif global de développer des méthodes économiques et robustes pour étudier la variabilité de la réponse statique ou dynamique des structures composites modélisées par éléments finis, prenant en compte les propriétés matériaux (modules d'élasticité, coefficients de Poisson, masses volumiques...) et physiques (épaisseurs et orientations des fibres) incertaines. Deux méthodes stochastiques : Certain Generalized Stresses Method (CGSM) et Modal Stability Procedure (MSP), sont développées. La méthode CGSM considère une hypothèse mécanique, les efforts généralisés sont supposés indépendants des paramètres incertains. Elle permet d'évaluer la variabilité de la réponse statique. La méthode MSP, proposée pour étudier la variabilité d'une structure en dynamique, est basée sur l'hypothèse considérant que les modes propres sont peu sensibles aux paramètres incertains. Les hypothèses mécaniques et une unique analyse par éléments finis permettent de construire un méta-modèle exploité dans une simulation de Monte Carlo. Le coût de calcul de ces méthodes stochastiques est donc réduit considérablement. De plus, elles présentent les avantages de ne pas limiter le nombre de paramètres incertains ou le niveau de variabilité d'entrée, et d'être compatibles avec tout code éléments finis standard. Quatre exemples académiques de plaque et coque composite sont traités avec la méthode CGSM, deux exemples académiques de plaque composite carrée et un exemple de plaque raidie sont traités avec la méthode MSP. La variabilité de la réponse statique (déplacement et critère de rupture) et dynamique (fréquence propre), soit la moyenne, l'écart-type, le coefficient de variation et la distribution, est évaluée. Les résultats statistiques obtenus par les méthodes proposées sont comparés avec ceux obtenus par une simulation de Monte Carlo directe, considérée comme la méthode de référence. La comparaison montre que les méthodes développées fournissent des résultats de bonne qualité et qu'elles sont très performantes en temps de calcul. Un indicateur d'erreur est également proposé, permettant de donner une estimation du niveau d'erreur des résultats obtenus par les méthodes CGSM ou MSP par rapport à la méthode de référence, sans réaliser un grand nombre d'analyses par éléments finis. / The manufacture of composite structures leads to a high variability of mechanical parameters. The objective of this work is to develop economic and robust methods to study the variability of the static or dynamic response of composite structures modeled by finite elements, taking into account uncertain material (elastic moduli, Poisson's ratios, densities... ) and physical (thicknesses and fiber orientations) properties. Two methods are developed: the Certain Generalized Stresses Method (CGSM) and the Modal Stability Procedure (MSP). The CGSM considers a mechanical assumption, the generalized stresses are assumed to be independant of uncertain parameters. lt allows to evaluate the variability of static response. The MSP, proposed to study the variability of structures in dynamics, is based on the assumption that the modes shapes are insensitive to uncertain parameters. These mechanical assumptions and only one fïnite element analysis allow to construct a metamodel used in a Monte Carlo simulation. As a result, the computational cost is reduced considerably. Moreover, they are not limited by the number of considered parameters or the level of input variability, and are compatible with standard finite element software. Four academic examples of composite plate and shell are treated with the CGSM, while two academic examples of composite square plate and an example of stiffened plate are studied by the MSP. The variability of static response (displacement and failure criterion) and dynamic response (natural frequency), namely mean value, standard deviation, coefficient of variation and distribution, is evaluated. The results obtained by the proposed methods are compared with those obtained by the direct Monte Carlo simulation, considered as the reference method. The comparison shows that the proposed methods provide quite accurate results and highlights their high computational efficiency. An error indicator is also proposed, which allows to provide an estimation of the error level of the results obtained by the CGSM or MSP compared to the reference method, without performing a large number of finite element analyses.

The aim of this diploma thesis is a selection of the best current account. This thesis describes how the bank system functions in the Czech Republic. The characteristics of the banking institutions which provide current accounts and the characteristics of the compared current accounts are presented in this thesis. The Monte Carlo method and the methods multi-criteria assessment of variants are described theoretically. The selection is based on the methods of the operational research. The first part of the practical chapter focuses on the Monte Carlo method which selects the best account by charges for different types of clients. The methods of multi-criteria assessment of variants, specifically methods TOPSIS and PROMETHEE, are also applied. The calculations are carried out by add-in SANNA for MS Excel. In conclusion, both approaches to solution are compared and the current account according to the preferences of each client type is recommended.

The thesis deals with the project of the construction of the Temelin Nuclear Power Plant, as an optimal new production source of electrical energy to ensure the majority of the coverage of the growing electric consumption in the Czech Republic and to ensure the state energy independence in the future. The main part of the thesis is dedicated to the methods of strategic investments evaluation. Emphasis is placed on the dynamic methods, which are also working with the factors of time and risk, which are relevant in this case, because the project time horizon is 70 years. The investment project is evaluated from the perspective of owners, as well as from the overall perspective of owners and creditors. The Monte-Carlo method was implemented into the model to support the decision-making process and to move closer to reality. The method assigns the relevant distribution division to the model input values. The output is the set of available values, and the probability of their occurrence. The main thesis objective lies in the decision of the decision maker with a neutral attitude to risk, whether accept the investment or not.