At the onset of the "Big Data" age, we are faced with ubiquitous data in various forms and with various characteristics, such as noise, high dimensionality, autocorrelation, and so on. The question of how to obtain accurate and computationally efficient estimates from such data is one that has stoked the interest of many researchers. This dissertation mainly concentrates on two general problem areas: inference for high-dimensional and noisy data, and estimation of the steady-state mean for univariate data generated by computer simulation experiments. We develop and evaluate three separate sequential algorithms for the two topics. One major advantage of sequential algorithms is that they allow for careful experimental adjustments as sampling proceeds. Unlike one-step sampling plans, sequential algorithms adapt to different situations arising from the ongoing sampling; this makes these procedures efficacious as problems become more complicated and more-delicate requirements need to be satisfied. We will elaborate on each research topic in the following discussion. Concerning the first topic, our goal is to develop a robust graphical model for noisy data in a high-dimensional setting. Under a Gaussian distributional assumption, the estimation of undirected Gaussian graphs is equivalent to the estimation of inverse covariance matrices. Particular interest has focused upon estimating a sparse inverse covariance matrix to reveal insight on the data as suggested by the principle of parsimony. For estimation with high-dimensional data, the influence of anomalous observations becomes severe as the dimensionality increases. To address this problem, we propose a robust estimation procedure for the Gaussian graphical model based on the Integrated Squared Error (ISE) criterion. The robustness result is obtained by using ISE as a nonparametric criterion for seeking the largest portion of the data that "matches" the model. Moreover, an l₁-type regularization is applied to encourage sparse estimation. To address the non-convexity of the objective function, we develop a sequential algorithm in the spirit of a majorization-minimization scheme. We summarize the results of Monte Carlo experiments supporting the conclusion that our estimator of the inverse covariance matrix converges weakly (i.e., in probability) to the latter matrix as the sample size grows large. The performance of the proposed method is compared with that of several existing approaches through numerical simulations. We further demonstrate the strength of our method with applications in genetic network inference and financial portfolio optimization. The second topic consists of two parts, and both concern the computation of point and confidence interval (CI) estimators for the mean µ of a stationary discrete-time univariate stochastic process X \equiv \{X_i: i=1,2,...} generated by a simulation experiment. The point estimation is relatively easy when the underlying system starts in steady state; but the traditional way of calculating CIs usually fails since the data encountered in simulation output are typically serially correlated. We propose two distinct sequential procedures that each yield a CI for µ with user-specified reliability and absolute or relative precision. The first sequential procedure is based on variance estimators computed from standardized time series applied to nonoverlapping batches of observations, and it is characterized by its simplicity relative to methods based on batch means and its ability to deliver CIs for the variance parameter of the output process (i.e., the sum of covariances at all lags). The second procedure is the first sequential algorithm that uses overlapping variance estimators to construct asymptotically valid CI estimators for the steady-state mean based on standardized time series. The advantage of this procedure is that compared with other popular procedures for steady-state simulation analysis, the second procedure yields significant reduction both in the variability of its CI estimator and in the sample size needed to satisfy the precision requirement. The effectiveness of both procedures is evaluated via comparisons with state-of-the-art methods based on batch means under a series of experimental settings: the M/M/1 waiting-time process with 90% traffic intensity; the M/H_2/1 waiting-time process with 80% traffic intensity; the M/M/1/LIFO waiting-time process with 80% traffic intensity; and an AR(1)-to-Pareto (ARTOP) process. We find that the new procedures perform comparatively well in terms of their average required sample sizes as well as the coverage and average half-length of their delivered CIs.

In this thesis, we first show that the performance of ranking and selection (R&S) procedures in steady-state simulations depends highly on the quality of the variance estimates that are used. We study the performance of R&S procedures using three variance estimators --- overlapping area, overlapping Cramer--von Mises, and overlapping modified jackknifed Durbin--Watson estimators --- that show better long-run performance than other estimators previously used in conjunction with R&S procedures for steady-state simulations. We devote additional study to the development of the new overlapping modified jackknifed Durbin--Watson estimator and demonstrate some of its useful properties. Next, we consider the problem of finding the best simulated system under a primary performance measure, while also satisfying stochastic constraints on secondary performance measures, known as constrained ranking and selection. We first present a new framework that allows certain systems to become dormant, halting sampling for those systems as the procedure continues. We also develop general procedures for constrained R&S that guarantee a nominal probability of correct selection, under any number of constraints and correlation across systems. In addition, we address new topics critical to efficiency of the these procedures, namely the allocation of error between feasibility check and selection, the use of common random numbers, and the cost of switching between simulated systems.

<p> A new computer executive system for the steady state simulation of chemical processes has been developed which combines modular (GEMCS) approach with the simultaneous linear equation (SYMBØL) approach to simulation. In the combined system, a GEMCS simulation, using non-linear models, is used to generate the coefficients for the set of linear equations describing the process. This linear system of equations may also include the constraints on the process which dictate the operating conditions for the actual process. The solution of the linear equations then provide new operating conditions (feed flowrates together with the component flowrates in the recycle streams) for the modular simulation, which in turn provides new coefficients; etc. This iterative procedure is automatically continued until the system is converged to the desired point. </p> <p> A modular simulation for an actual Naphtha Reforming Plant has also been achieved and it was used as a test case to demonstrate the use and effectiveness of this new executive system. In the course of developing this simulation, the application of a method for correcting plant data was demonstrated. This is the first real application of this method to be reported in the current literature. </p> / Thesis / Master of Engineering (ME)

[pt] O tema abordado neste trabalho é a observação e análise, em regime permanente e dinâmico, da ocorrência de um fenômeno que já foi observado em condições reais de operação do sistema elétrico brasileiro, que é a relação oposta entre a tensão de excitação de geradores e compensadores síncronos e a tensão controlada. Nessas situações, a capacidade nominal de um gerador / compensador síncrono, por exemplo, não seria útil para manter a tensão controlada. Em virtude da relação oposta, um aumento na excitação da máquina abaixaria a tensão controlada. O controle automático continuaria agindo, abaixando ainda mais a tensão. Este mecanismo pode levar o sistema ao colapso. A abordagem do problema baseou-se na verificação do comportamento do gerador / compensador como dispositivo de controle de tensão, em regime permanente e dinâmico, ante as diversas situações normais de um sistema elétrico, tais como variações do valor da tensão de referência (tensão controlada) e de carga. A análise em regime permanente utilizou um algoritmo de fluxo de carga, enquanto a análise em regime dinâmico utilizou a simulação no domínio do tempo. A real existência do fenômeno foi comprovada através de várias destas análises, mostrando inclusive a mudança da região de operação em algumas delas. Em outros casos, os resultados da análise em regime permanente não coincidiram com os da análise em regime dinâmico. / [en] The aim of this work is to investigate, in steady state and dynamic performance, the phenomenon of the opposite relationship, already observed at real operation conditions of the Brazilian Electric System, between generators and synchronous compensators excitation voltage and the controlled one. In these situations, the generator / synchronous compensator nominal capacity, for example, would not be useful to keep the voltage controlled. Due the opposite relationship, an increase in the excitation voltage would reduce the controlled voltage. The automatic control would keep acting and reducing more the voltage. This mechanism can lead the system to the collapse. The study of this problem was based in the generator / compensator behavior as a control voltage device, in steady- state and dynamic performance, front of several operation situations of electric power system, like reference voltage (controlled voltage) variation and load changing. The steady state analysis used a load flow algorithm, while the time domain simulation was utilized for the dynamic performance analysis. The real existence of the phenomenon was verified through these analyses, emphasizing the operation region changing in some of them. In other cases, the analyses results in the steady-state were different of the dynamic performance results.

[pt] O presente trabalho trata do projeto, modelagem e simulação de uma bancada de testes, com a finalidade de mensurar o desempenho de componentes de refrigeração, para ciclos por compressão de vapor. A configuração da bancada foi concebida para oferecer flexibilidade na escolha dos componentes e estabilidade durante a realização dos experimentos. Um modelo matemático foi desenvolvido e implementado a partir da aplicação dos princípios de conservação da massa e energia para cada um dos volumes de controle dos componentes da bancada. A modelagem considera também as equações de troca de calor, propriedades termodinâmicas e modelos físicos dos componentes, para prever condições de operação da bancada operando em regime permanente. Para a modelagem computacional utilizou-se o software EES-Engineering Equation Solver. O modelo resultante também foi utilizado no pré-dimensionamento dos subsistemas da bancada e na simulação de parâmetros característicos, a fim de delimitar a faixa de capacidades, bem como as variáveis de entrada e saída, necessárias ao controle do sistema. O dimensionamento e a seleção dos equipamentos e componentes foram efetuados com base no refrigerante R134a, como fluido de trabalho. No entanto, o modelo é compatível com a utilização de outros refrigerantes, tais como o R22, R407C, R410A, e outras misturas. O trabalho inclui um modelo para simular o comportamento em regime transiente do tanque inercial térmico com o intuito de estudar a estabilidade e o controle do sistema em regime permanente. A simulação demonstrou que a configuração da bancada garantirá estabilidade e reprodutibilidade dos experimentos nas eventuais variações das condições externas, diferindo, portanto, de grande parte dos sistemas de refrigeração convencionais. / [en] The present work deals with the design, modeling and simulation of a test bench used in evaluating the performance of components of vapor compression refrigeration systems. The mathematical model was developed and implemented based on the application of the conservation of mass and energy principles to each of the control volumes, heat transfer equations, thermodynamic properties for refrigerant and water and existing physical models for the system components were also applied. The software EES-Engineering Equation Solver was used as the computational platform. The resulting simulation program was initially used to size the test bench subsystems and to simulate characteristic parameters in order to determine the range of capacities and input and output variables. Refrigerant R-134a was the working substance considered when sizing and choosing system components, but the model can also be used with R22 refrigerants as well as R410A, R407C and others refrigerant mixtures. The work also includes a transient analysis of the thermal inertial tank with focus on the study of the operational control for the system being operated at steady state. The simulation showed that the layout of the test bench will warrant stability and reproducibility of experiments for eventual changes in the external condition, thus differing from typical conventional vapor compression refrigeration systems.

[pt] O problema tratado nesta dissertação é a relação oposta entre a tensão de excitação de geradores e compensadores síncronos e a tensão controlada, quando o sistema de transmissão da área encontra-se muito carregado. Neste caso, a capacidade nominal de um gerador / compensador não seria útil para manter a tensão controlada. Devido à relação oposta, uma maior excitação da máquina iria abaixar a tensão controlada. O controle automático iria continuar agindo, abaixando ainda mais a tensão. Este mecanismo pode levar o sistema ao colapso e foi verificado em ponto de operação real do sistema brasileiro. Esse fenômeno ocorre quando a injeção de potência na rede de transmissão ou distribuição é elevada. Com o advento da geração distribuída, co-geração e produtores independentes, usualmente conectados à rede existente em níveis de tensão mais baixas, têm-se observado ocorrências do fenômeno. O objetivo do trabalho é então entender melhor as situações operativas reais que levam à ocorrência do fenômeno, principalmente quando existem vários equipamentos de controle de tensão ao redor do gerador em análise. A abordagem do problema baseou-se na verificação do comportamento do gerador / compensador como dispositivo de controle de tensão, no domínio do tempo e em regime permanente. Avaliaram-se as ações de controle do mesmo a partir de sete tipos de análise distintas para pontos de operação na região normal e anormal da curva SV. A real existência do fenômeno foi comprovada através de algumas destas análises. Porém, conclui-se que nem todas as formas de análise no domínio do tempo fazem uma avaliação completa do fenômeno. Dentre essas, estão a análise dinâmica agregada e a análise dinâmica agregada sob influência dos equipamentos de controle de tensão, que para pontos de operação na região anormal da curva SV não responderam em concordância com os outros tipos de análise. / [en] The problem addressed in this research is the opposite relationship between the , synchronous generator / compensator excitation voltage and the controlled voltage when nearby network is heavily loaded. In this situation, the nominal capacity of a generator / compensator would not keep the voltage controlled. Due to the opposite relationship, the higher the excitation voltage the lower is the controlled voltage. So, the automatic control would continue acting lowering the voltage. This mechanism, verified in a real operational point of the Brazilian Electric System, can lead the system to collapse. This phenomenon occurs when the power injection into the network is high. It is prone to occur in the new scenario of distributed generation connected to already existing low voltage networks. The objective of this work is to understand the actual operative situations that lead to the occurrence of the phenomenon, mainly when there are several voltage control devices nearby the generator. The analysis of the problem was based on the verification of the generator behaviour as a voltage control device, in time domain simulation and in steady state. The control actions were evaluated from seven different ways for operating points in the normal and the abnormal region of the SV curve. The actual existence of the phenomenon was proven through some of these analyses. However, some of the time domain simulations did not evaluated the phenomenon completely. Among them, the aggregated dynamic analysis and the aggregated dynamic analysis under influence of other voltage control devices have not got the expected responses for the abnormal region of the SV curve, in comparison with other analysis.