Dimensionality hyper-reduction and machine learning for dynamical systems with varying parameters

Thesis (Ph.D.)--University of Washington, 2016-08

Modélisation des machines par circuits couplés pour l’observation des courants de barres au rotor

Ce travail présente une modélisation rapide d’ordre élévé capable de modéliser une configuration rotorique en cage complète ou en grille, de reproduire les courants de barre et tenir compte des harmoniques d’espace. Le modèle utilise une approche combinée d’éléments finis avec les circuits-couplés. En effet, le calcul des inductances est réalisé avec les éléments finis, ce qui confère une précision avancée au modèle. Cette méthode offre un gain important en temps de calcul sur les éléments finis pour des simulations transitoires. Deux outils de simulation sont développés, un dans le domaine du temps pour des résolutions dynamiques et un autre dans le domaine des phaseurs dont une application sur des tests de réponse en fréquence à l’arrêt (SSFR) est également présentée. La méthode de construction du modèle est décrite en détail de même que la procédure de modélisation de la cage du rotor. Le modèle est validé par l’étude de machines synchrones: une machine de laboratoire de 5.4 KVA et un grand alternateur de 109 MVA dont les mesures expérimentales sont comparées aux résultats de simulation du modèle pour des essais tels que des tests à vide, des courts-circuits triphasés, biphasés et un test en charge.

Dynamic Resource Allocation in Wireless Heterogeneous Networks

Deployment of low power basestations within cellular networks can potentially increase both capacity and coverage. However, such deployments require efficient resource allocation schemes for managing interference from the low power and macro basestations that are located within each other’s transmission range. In this dissertation, we propose novel and efficient dynamic resource allocation algorithms in the frequency, time and space domains. We show that the proposed algorithms perform better than the current state-of-art resource management algorithms. In the first part of the dissertation, we propose an interference management solution in the frequency domain. We introduce a distributed frequency allocation scheme that shares frequencies between macro and low power pico basestations, and guarantees a minimum average throughput to users. The scheme seeks to minimize the total number of frequencies needed to honor the minimum throughput requirements. We evaluate our scheme using detailed simulations and show that it performs on par with the centralized optimum allocation. Moreover, our proposed scheme outperforms a static frequency reuse scheme and the centralized optimal partitioning between the macro and picos. In the second part of the dissertation, we propose a time domain solution to the interference problem. We consider the problem of maximizing the alpha-fairness utility over heterogeneous wireless networks (HetNets) by jointly optimizing user association, wherein each user is associated to any one transmission point (TP) in the network, and activation fractions of all TPs. Activation fraction of a TP is the fraction of the frame duration for which it is active, and together these fractions influence the interference seen in the network. To address this joint optimization problem which we show is NP-hard, we propose an alternating optimization based approach wherein the activation fractions and the user association are optimized in an alternating manner. The subproblem of determining the optimal activation fractions is solved using a provably convergent auxiliary function method. On the other hand, the subproblem of determining the user association is solved via a simple combinatorial algorithm. Meaningful performance guarantees are derived in either case. Simulation results over a practical HetNet topology reveal the superior performance of the proposed algorithms and underscore the significant benefits of the joint optimization. In the final part of the dissertation, we propose a space domain solution to the interference problem. We consider the problem of maximizing system utility by optimizing over the set of user and TP pairs in each subframe, where each user can be served by multiple TPs. To address this optimization problem which is NP-hard, we propose a solution scheme based on difference of submodular function optimization approach. We evaluate our scheme using detailed simulations and show that it performs on par with a much more computationally demanding difference of convex function optimization scheme. Moreover, the proposed scheme performs within a reasonable percentage of the optimal solution. We further demonstrate the advantage of the proposed scheme by studying its performance with variation in different network topology parameters.

Application of the latency insertion method (LIM) to the analysis of large circuit interconnect networks

The focus of this research is to explore the applications of the finite difference formulation based on the latency insertion method (LIM) to the analysis of circuit interconnects. Special attention is devoted to addressing the issues that arise in very large networks such as on-chip signal and power distribution networks. We demonstrate that the LIM has the power and flexibility to handle various types of analysis required at different stages of circuit design. The LIM is particularly suitable for simulations of very large scale linear networks and can significantly outperform conventional circuit solvers (such as SPICE).

A test method for analysing disturbed ethernet data streams

Ethernet connections, which are widely used in many computer networks, can suffer from electromagnetic interference. Typically, a degradation of the data transmission rate can be perceived as electromagnetic disturbances lead to corruption of data frames on the network media. In this paper a software-based measuring method is presented, which allows a direct assessment of the effects on the link layer. The results can directly be linked to the physical interaction without the influence of software related effects on higher protocol layers. This gives a simple tool for a quantitative analysis of the disturbance of an Ethernet connection based on time domain data. An example is shown, how the data can be used for further investigation of mechanisms and detection of intentional electromagnetic attacks. © 2015 Author(s).

Real-time anomaly detection of gamma-ray bursts for the Cherenkov Telescope Array using deep learning

The Cherenkov Telescope Array (CTA) will be the next-generation ground-based observatory to study the universe in the very-high-energy domain. The observatory will rely on a Science Alert Generation (SAG) system to analyze the real-time data from the telescopes and generate science alerts. The SAG system will play a crucial role in the search and follow-up of transients from external alerts, enabling multi-wavelength and multi-messenger collaborations. It will maximize the potential for the detection of the rarest phenomena, such as gamma-ray bursts (GRBs), which are the science case for this study. This study presents an anomaly detection method based on deep learning for detecting gamma-ray burst events in real-time. The performance of the proposed method is evaluated and compared against the Li&Ma standard technique in two use cases of serendipitous discoveries and follow-up observations, using short exposure times. The method shows promising results in detecting GRBs and is flexible enough to allow real-time search for transient events on multiple time scales. The method does not assume background nor source models and doe not require a minimum number of photon counts to perform analysis, making it well-suited for real-time analysis. Future improvements involve further tests, relaxing some of the assumptions made in this study as well as post-trials correction of the detection significance. Moreover, the ability to detect other transient classes in different scenarios must be investigated for completeness. The system can be integrated within the SAG system of CTA and deployed on the onsite computing clusters. This would provide valuable insights into the method's performance in a real-world setting and be another valuable tool for discovering new transient events in real-time. Overall, this study makes a significant contribution to the field of astrophysics by demonstrating the effectiveness of deep learning-based anomaly detection techniques for real-time source detection in gamma-ray astronomy.

Calibração da Reflectometria no Domínio do Tempo (TDR) para a estimativa da concentração da solução no solo

Time Domain Reflectometry (TDR) is a reliable method for in-situ measurements of the humidity and the solution concentration at the same soil volume. Accurate interpretation of electrical conductivity (and soil humidity) measurements may require a specific calibration curve. The primary goal of this work was to establish a calibration procedure for using TDR to estimate potassium nitrate concentrations (KNO3) in soil solution. An equation relating the electrical conductivity measured by TDR and KNO3 concentration was established enabling the use of TDR technique to estimate soil water content and nitrate concentration for efficient fertigation management.

Análise quantitativa da distribuição de jogadores de futebol em campo durante jogos oficiais

Universidade Estadual de Campinas. Faculdade de Educação Física

Controle autonomico da frequencia cardiaca de adolescentes em treinamento

Universidade Estadual de Campinas . Faculdade de Educação Física

a-SiC : H anti-resonant layer ARROW waveguides

Over the last decades, anti-resonant reflecting optical waveguides (ARROW) have been used in different integrated optics applications. In this type of waveguide, light confinement is partially achieved through an anti-resonant reflection. In this work, the simulation, fabrication and characterization of ARROW waveguides using dielectric films deposited by a plasma-enhanced chemical vapor deposition (PECVD) technique, at low temperatures(similar to 300 degrees C), are presented. Silicon oxynitride (SiO(x)N(y)) films were used as core and second cladding layers and amorphous hydrogenated silicon carbide(a-SiC:H) films as first cladding layer. Furthermore, numerical simulations were performed using homemade routines based on two computational methods: the transfer matrix method (TMM) for the determination of the optimum thickness of the Fabry-Perot layers; and the non-uniform finite difference method (NU-FDM) for 2D design and determination of the maximum width that yields single-mode operation. The utilization of a silicon carbide anti-resonant layer resulted in low optical attenuations, which is due to the high refractive index difference between the core and this layer. Finally, for comparison purposes, optical waveguides using titanium oxide (TiO(2)) as the first ARROW layer were also fabricated and characterized.

Performance of a multi-level TDR system exposed to tropical field conditions - A time-space comparison with tensiometry

Time-domain reflectometry (TDR) is an important technique to obtain series of soil water content measurements in the field. Diode-segmented probes represent an improvement in TDR applicability, allowing measurements of the soil water content profile with a single probe. In this paper we explore an extensive soil water content dataset obtained by tensiometry and TDR from internal drainage experiments in two consecutive years in a tropical soil in Brazil. Comparisons between the variation patterns of the water content estimated by both methods exhibited evidences of deterioration of the TDR system during this two year period at field conditions. The results showed consistency in the variation pattern for the tensiometry data, whereas TDR estimates were inconsistent, with sensitivity decreasing over time. This suggests that difficulties may arise for the long-term use of this TDR system under tropical field conditions. (c) 2008 Elsevier B.V. All rights reserved.

Computationally efficient solution techniques for adsorption problems involving steep gradients in bidisperse particles

A piecewise uniform fitted mesh method turns out to be sufficient for the solution of a surprisingly wide variety of singularly perturbed problems involving steep gradients. The technique is applied to a model of adsorption in bidisperse solids for which two fitted mesh techniques, a fitted-mesh finite difference method (FMFDM) and fitted mesh collocation method (FMCM) are presented. A combination (FMCMD) of FMCM and the DASSL integration package is found to be most effective in solving the problems. Numerical solutions (FMFDM and FMCMD) were found to match the analytical solution when the adsorption isotherm is linear, even under conditions involving steep gradients for which global collocation fails. In particular, FMCMD is highly efficient for macropore diffusion control or micropore diffusion control. These techniques are simple and there is no limit on the range of the parameters. The techniques can be applied to a variety of adsorption and desorption problems in bidisperse solids with non-linear isotherm and for arbitrary particle geometry.

A two-dimensional analytical solution of groundwater responses to tidal loading in an estuary and ocean

Previous studies on tidal dynamics of coastal aquifers have focussed on the inland propagation of oceanic tides in the cross-shore direction, a configuration that is essentially one-dimensional. Aquifers at natural coasts can also be influenced by tidal waves in nearby estuaries, resulting in a more complex behaviour of head fluctuations in the aquifers. We present an analytical solution to the two-dimensional depth-averaged groundwater flow equation for a semi-infinite aquifer subject to oscillating head conditions at the boundaries. The solution describes the tidal dynamics of a coastal aquifer that is adjacent to a cross-shore estuary. Both the effects of oceanic and estuarine tides on the aquifer are included in the solution. The analytical prediction of the head fluctuations is verified by comparison with numerical solutions computed using a standard finite-difference method. An essential feature of the present analytical solution is the interaction between the cross- and along-shore tidal waves in the aquifer area near the estuary's entry. As the distance from the estuary or coastline increases, the wave interaction is weakened and the aquifer response is reduced, respectively, to the one-dimensional solution for oceanic tides or the solution of Sun (Sun H. A two-dimensional analytical solution of groundwater response to tidal loading in an estuary, Water Resour Res 1997;33:1429-35) for two-dimensional non-interacting tidal waves. (C) 2000 Elsevier Science Ltd. All rights reserved.

Numerical modelling of double diffusion driven reactive flow transport in deformable fluid-saturated porous media with particular consideration of temperature-dependent chemical reaction rates

Numerical methods ave used to solve double diffusion driven reactive flow transport problems in deformable fluid-saturated porous media. in particular, thp temperature dependent reaction rate in the non-equilibrium chemical reactions is considered. A general numerical solution method, which is a combination of the finite difference method in FLAG and the finite element method in FIDAP, to solve the fully coupled problem involving material deformation, pore-fluid flow, heat transfer and species transport/chemical reactions in deformable fluid-saturated porous media has been developed The coupled problem is divided into two subproblems which are solved interactively until the convergence requirement is met. Owing to the approximate nature of the numerical method, if is essential to justify the numerical solutions through some kind of theoretical analysis. This has been highlighted in this paper The related numerical results, which are justified by the theoretical analysis, have demonstrated that the proposed solution method is useful for and applicable to a wide range of fully coupled problems in the field of science and engineering.

Theoretical and numerical analyses of convective instability in porous media with temperature-dependent viscosity

Exact analytical solutions of the critical Rayleigh numbers have been obtained for a hydrothermal system consisting of a horizontal porous layer with temperature-dependent viscosity. The boundary conditions considered are constant temperature and zero vertical Darcy velocity at both the top and bottom of the layer. Not only can the derived analytical solutions be readily used to examine the effect of the temperature-dependent viscosity on the temperature-gradient driven convective flow, but also they can be used to validate the numerical methods such as the finite-element method and finite-difference method for dealing with the same kind of problem. The related analytical and numerical results demonstrated that the temperature-dependent viscosity destabilizes the temperature-gradient driven convective flow and therefore, may affect the ore body formation and mineralization in the upper crust of the Earth. Copyright (C) 2003 John Wiley Sons, Ltd.