Generalized linear mixture models (GLMM) are widely used in statistical applications to model count and binary data. We consider the problem of nonparametric likelihood estimation of mixing distributions in GLMM's with multiple random effects. The log-likelihood to be maximized has the general form l(G)=Σi log∫f(yi,γ) dG(γ) where f(.,γ) is a parametric family of component densities, yi is the ith observed response dependent variable, and G is a mixing distribution function of the random effects vector γ defined on Ω. The literature presents many algorithms for maximum likelihood estimation (MLE) of G in the univariate random effect case such as the EM algorithm (Laird, 1978), the intra-simplex direction method, ISDM (Lesperance and Kalbfleish, 1992), and vertex exchange method, VEM (Bohning, 1985). In this dissertation, the constrained Newton method (CNM) in Wang (2007), which fits GLMM's with random intercepts only, is extended to fit clustered datasets with multiple random effects. Owing to the general equivalence theorem from the geometry of mixture likelihoods (see Lindsay, 1995), many NPMLE algorithms including CNM and ISDM maximize the directional derivative of the log-likelihood to add potential support points to the mixing distribution G. Our method, Direct Search Directional Derivative (DSDD), uses a directional search method to find local maxima of the multi-dimensional directional derivative function. The DSDD's performance is investigated in GLMM where f is a Bernoulli or Poisson distribution function. The algorithm is also extended to cover GLMM's with zero-inflated data. Goodness-of-fit (GOF) and selection methods for mixed models have been developed in the literature, however their application in models with nonparametric random effects distributions is vague and ad-hoc. Some popular measures such as the Deviance Information Criteria (DIC), conditional Akaike Information Criteria (cAIC) and R2 statistics are potentially useful in this context. Additionally, some cross-validation goodness-of-fit methods popular in Bayesian applications, such as the conditional predictive ordinate (CPO) and numerical posterior predictive checks, can be applied with some minor modifications to suit the non-Bayesian approach. / Graduate / 0463 / rabihsaab@gmail.com

Photovoltaic manufacturing is material intensive with the cost of crystalline silicon wafer, used as the substrate, representing 40% to 60% of the solar cell cost. Consequently, there is a growing trend to reduce the silicon wafer thickness leading to new technical challenges related to manufacturing. Specifically, wafer breakage during handling and/or transfer is a significant issue. Therefore improved methods for breakage-free handling are needed to address this problem. An important pre-requisite for realizing such methods is the need for fundamental understanding of the effect of handling device variables on the deformation, stresses, and fracture of crystalline silicon wafers. This knowledge is lacking for wafer handling devices including the Bernoulli gripper, which is an air flow nozzle based device. A computational fluid dynamics model of the air flow generated by a Bernoulli gripper has been developed. This model predicts the air flow, pressure distribution and lifting force generated by the gripper. For thin silicon wafers, the fluid model is combined with a finite element model to analyze the effects of wafer flexibility on the equilibrium pressure distribution, lifting force and handling stresses. The effect of wafer flexibility on the air pressure distribution is found to be increasingly significant at higher air flow rates. The model yields considerable insight into the relative effects of air flow induced vacuum and the direct impingement of air on the wafer on the air pressure distribution, lifting force, and handling stress. The latter effect is found to be especially significant when the wafer deformation is large. In addition to silicon wafers, the model can also be used to determine the lifting force and handling stress produced in other flexible materials. Finally, a systematic approach for the analysis of the total stress state (handling plus residual stresses) produced in crystalline silicon wafers and its impact on wafer breakage during handling is presented. Results confirm the capability of the approach to predict wafer breakage during handling given the crack size, location and fracture toughness. This methodology is general and can be applied to other thin wafer handling devices besides the Bernoulli gripper.

This MA thesis focuses on the aerodynamic measurement in the teaching of physics using the Leybold Didactic didactic set. The thesis consists of instructions for solving 4 basic problems using the Leybold set; every problem comprises theoretical instructions for students and an experimental verification of the problem. The thesis should contribute to wider use of experimental teaching of physics in schools; also, it is supposed to serve teachers and students as a source of interesting laboratory assignments verifying aerodynamic laws and quantities. After minor adjustments, the thesis can be used in all types of schools, from elementary physics at primary schools to advanced physics at technical universities.

Submitted by Automa??o e Estat?stica (sst@bczm.ufrn.br) on 2017-04-03T22:46:24Z No. of bitstreams: 1 TitoLivioDaCunhaLopes_DISSERT.pdf: 830141 bytes, checksum: a867c27dace025040774c75e8896b7e2 (MD5) / Approved for entry into archive by Arlan Eloi Leite Silva (eloihistoriador@yahoo.com.br) on 2017-04-11T21:20:49Z (GMT) No. of bitstreams: 1 TitoLivioDaCunhaLopes_DISSERT.pdf: 830141 bytes, checksum: a867c27dace025040774c75e8896b7e2 (MD5) / Made available in DSpace on 2017-04-11T21:20:49Z (GMT). No. of bitstreams: 1 TitoLivioDaCunhaLopes_DISSERT.pdf: 830141 bytes, checksum: a867c27dace025040774c75e8896b7e2 (MD5) Previous issue date: 2016-11-28 / Conselho Nacional de Desenvolvimento Cient?fico e Tecnol?gico (CNPq) / Coordena??o de Aperfei?oamento de Pessoal de N?vel Superior (CAPES) / Modelos para s?ries temporais de valores inteiros t?m se destacado devido a vasta possibilidade de aplica??o. Modelos para controle estat?stico de processos, para dados econ?micos e, atualmente, para a sequ?ncia estrutural dos ?cidos desoxirribonucleicos (DNA), s?o exemplos de importantes aplica??es. Este trabalho est? dividido em dois cap?tulos independentes. A primeira parte do trabalho diz respeito a modelagem de dados bin?rios autocorrelacionados. Neste contexto, uma nova classe de modelos foi proposta, baseado em operadores thinning, denominada processo Bernoulli autorregressivo de ordem p[BeAr(p)] similar ao modelo cl?ssico AR(p). Em particular, o modelo BeAr(1) foi estudado e v?rias propriedades foram estabelecidas, tr?s m?todos de estima??o foram propostos para o modelo, inclusive foi estabelecida a distribui??o assint?tica dos estimadores pelo m?todo de m?nimos quadrados condicionais e os elementos da matriz de informa??o de Fisher. Al?m das simula??es, aplica??es foram feitas em dados reais de precipita??o, ocasi?o em que os modelos BeAr(1) e BeAr(2) foram indicados para modelagem. Na segunda parte do trabalho, novos modelos foram estudados ao propor a fam?lia de distribui??es de s?ries de pot?ncia generalizada com par?metro inflador (IGPSD) para o processo de inova??o do modelo INAR(1). As principais propriedades do processo foram estabelecidas, tais como a m?dia, vari?ncia, autocorrela??o e probabilidade de transi??o. Os m?todos de estima??o por Yule-Walker e m?xima verossimilhan?a condicional foram utilizados para estimar os par?metros dos modelos. Dois casos particulares do modelo INAR$(1)$ com inova??o IGPSD foram estudados, denominados de IPoINAR(1) e IGeoINAR(1). Por fim, na aplica??o a dados reais, observou-se um bom desempenho do novo modelo proposto. / Models for time series of integer values have stood out because of the vast possibility of application. Models for statistical process control, for economic data and currently for the structural sequence of deoxyribonucleic acids (DNA) are examples of important applications. This work is divided into two independent parts. The first part of the work concerns the modeling of autocorrelated binary data. In this context, a new class of models has been proposed, based on thinning operators called Bernoulli autoregressive process of order p [BeAr(p)] similar to the classical model AR(p). In particular, BeAr(1) model was studied, various properties of three estimation methods have been proposed for the model, including the asymptotic distribution of the estimators by the conditional least squares method, and the elements of the Fisher information matrix. In addition to the simulations, applications were made on real data of precipitation, at which models BeAr(1) and BeAr(2) were given to modeling. In the second part of the work, new models were studied to propose the family of inflated-parameter generalized power series distributions (IGPSD) to the innovation process INAR(1) model. The main properties of the process have been established, such as the mean, variance, autocorrelation and transition probability. The estimation methods for Yule-Walker and conditional maximum likelihood were used to estimate the parameters of the models. Two particular cases of model INAR(1) with IGPSD innovation process were studied, called IPoINAR(1) and IGeoINAR(1). Applications to real data showed a good performance of the new model proposed.

Rotating Euler-Bernoulli beams and non-homogeneous Timoshenko beams are widely used to model important engineering structures. Hence the vibration analyses of these beams are an important problem from a structural dynamics point of view. The governing differential equations of both these type of beams do not yield any simple closed form solutions, hence we look for the inverse problem approach in determining the beam property variations given certain solutions. Firstly, we look for a rotating beam, with pinned-free boundary conditions, whose eigenpair (frequency and mode-shape) is same as that of a uniform non-rotating beam for a particular mode. It is seen that for any given mode, there exists a flexural stiffness function (FSF) for which the ith mode eigenpair of a rotating beam with uniform mass distribution, is identical to that of a corresponding non-rotating beam with same length and mass distribution. Inserting these derived FSF's in a finite element code for a rotating pinned-free beam, the frequencies and mode shapes of a non-rotating pinned-free beam are obtained. For the first mode, a physically realistic equivalent rotating beam is possible, but for higher modes, the FSF has internal singularities. Strategies for addressing these singularities in the FSF for finite element analysis are provided. The proposed functions can be used as test functions for rotating beam codes and also for targeted destiffening of rotating beams. Secondly, we study the free vibration of rotating Euler-Bernoulli beams, under cantilever boundary condition. For certain polynomial variations of the mass per unit length and the flexural stiffness, there exists a fundamental closed form solution to the fourth order governing differential equation. It is found that there are an infinite number of rotating beams, with various mass per unit length variations and flexural stiffness distributions, which share the same fundamental frequency and mode shape. The derived flexural stiffness polynomial functions are used as test functions for rotating beam numerical codes. They are also used to design rotating cantilever beams which may be required to vibrate with a particular frequency. Thirdly, we study the free vibration of non-homogeneous Timoshenko beams, under fixed-fixed and fixed-hinged boundary conditions. For certain polynomial variations of the material mass density, elastic modulus and shear modulus, there exists a fundamental closed form solution to the coupled second order governing differential equations. It is found that there are an infinite number of non-homogeneous Timoshenko beams, with various material mass density, elastic modulus and shear modulus distributions, which share the same fundamental frequency and mode shape. They can be used to design non-homogeneous Timoshenko beams which may be required for certain engineering applications.

This thesis work provides an insight into the design of Force-amplifying Compliant Mechanisms (FaCMs) that are integrated with micromachined resonant accelerometers to increase their sensitivity. An FaCM, by mechanically amplifying the inertial force, enhances the shift in the resonance frequency of the beams used for sensing the acceleration whose effect causes an axial force on the beams. An extensive study on different configurations of resonators namely, single beam resonator, single-ended tuning fork (SETF), and double-ended tuning fork (DETF), is carried out to gain insights about their resonant behavior. The influence of the boundary conditions on the sensor’s sensitivity emerged from the study. We found that not only the force-amplification factor but also the multi-axial stiffness of the FaCM and proof-mass influence the resonance frequency of the resonator as well as the bandwidth of the modified sensor for certain configurations but not all. Thus, four lumped parameters were identified to quantify the effectiveness of an FaCM. These parameters determine the boundary condition of the sensing beams and also the forces and the moment transmitted to them. Also presented in this work is a computationally efficient model, called the Lumped Parameter Model (LPM) for evaluation of the sensitivity. An analytical expression for the frequency-shift of the sensing resonator beams is obtained by considering the FaCM stiffness parameters as well as the lumped stiffness of the suspension of the inertial mass. Various FaCMs are evaluated and compared to understand how the four lumped parameters influence the sensor’s sensitivity. The FaCMs are synthesized using topology optimization to maximize the net amplification factor with the volume constraint. One of the FaCMs outperforms the lever by a factor of six. Microfabrication of resonant accelerometer coupled with FaCM and comb-drive actuator is carried out using a silicon-on-insulator process. Finally, the selection map technique, a compliant mechanism redesign methodology is used for enhancing the amplification of FaCMs. This technique provides scope for further design improvement in FaCMs for given sensor specifications.

In der Natur und in technischen Anwendungen findet man viele dynamische System mit zeitlicher Verzögerung (delay),zum Beispiel die Mackey-Glass Gleichung, die als Beschreibung vom Aufbau der Blutzelle angewendet wird, und die Ikeda Gleichung, die ein Modell von einem optischen Resonator ist. Hier ist zeitliche Verzögerung τ eine Konstante, aber sie ist nicht immer konstant in der Natur und in der Praxis. Wie sieht das System aus und welche Stabilitätseigenschaften hat es, wenn die Verzögerung schwankt? In dieser Arbeit benutze ich einige einfache diskrete Abbildungen, um die resultierenden Effekte zu untersuchen.

No description available.

No description available.

<p>Multiple target tracking (MTT) is a major area that occurs in a variety of real world systems. The problem involves the detection and estimation of an unknown number of targets within a scenario space given a sequence of noisy, incomplete measurements. The classic approach to MTT performs data association between individual measurements, however, this step is a computationally complex problem. Recently, a series of algorithms based on Random Finite Set (RFS) theory, that do not require data association, have been introduced. This thesis addresses some of the main deficiencies involved with RFS methods and derives key extensions to improve them for use in real world systems.\\</p> <p>The first contribution is the Weight Partitioned PHD filter. It separates the Probability Hypothesis Density (PHD) surface into partitions that represent the individual state estimates both spatially and proportionally. The partitions are labeled and propagated over several time steps to form continuous track estimates. Multiple variants of the filter are presented. Next, the Multitarget Multi-Bernoulli (MeMBer) filter is extended to allow the tracking of manoeuvring targets. A model state variable is incorporated into the filter framework to estimate the probability of each motion model. The standard implementations are derived. Finally, a new linear variant of the Intensity filter (iFilter) is presented. A Gaussian Mixture approximation provides more computationally efficient implementation of the iFilter.</p> <p>Each of the new algorithms are validated on simulated data using standard multitarget tracking metrics. In each case, the methods improve on several aspects of multitarget tracking in the real world.</p> / Doctor of Engineering (DEng)