京都大学 / 新制・課程博士 / 博士(工学) / 甲第23873号 / 工博第4960号 / 新制||工||1775(附属図書館) / 京都大学大学院工学研究科建築学専攻 / (主査)教授 大崎 純, 教授 池田 芳樹, 准教授 藤田 皓平 / 学位規則第4条第1項該当 / Doctor of Philosophy (Engineering) / Kyoto University / DFAM

Over the past thirty years, extensive research has been conducted with the objective of reducing wind damage to structures. Wind tunnel simulations of wind loads have been the major source of building codes. However, a simple comparison of pressure coefficients measured in wind tunnel simulations with full-scale measurements show that the simulations, in general, underpredict extreme negative pressure coefficients. One obvious reason is the lack of consensus on wind tunnel simulation parameters. The wind in the atmospheric surface layer is highly turbulent. In simulating wind loads on structures, one needs to simulate the turbulent character besides satisfying geometric and dynamic similitudes. Some turbulence parameters that have been considered in many simulations include, turbulence intensities, integral length scales, surface roughness, and frequency spectrum. One problem with these parameters is that they are time varying in the atmospheric boundary layer and their averaged value, usually considered in the wind tunnel simulations, cannot be used to simulate pressure peaks. In this work, we show how wavelet analysis and time-scale representation can be used to establish an intermittency factor that characterizes energetic turbulence events in the atmospheric flows. Moreover, we relate these events to the occurrence of extreme negative peak pressures. / Ph. D.

A parametric identification procedure is proposed that combines the method of multiple scales and higher-order statistics to efficiently and accurately model nonlinear systems. A theoretical background for the method of multiple scales and higher-order statistics is given. Validation of the procedure is performed through applying it to numerical simulations of two nonlinear systems. The results show how the procedure can successfully characterize the system damping and nonlinearities and determine the corresponding parameters. The procedure is then applied to experimental measurements from two structural systems, a cantilevered beam and a three-beam frame. The results show that quadratic damping should be accounted for in both systems. Moreover, for the three-beam frame, the parametric excitation is much more important than the direct excitation. To show the flexibility of the procedure, numerical simulations of ship motion under parametric excitation are used to determine nonlinear parameters govening the relation between pitch, heave, and roll motions. The results show a high level of agreement between the numerical simulation and the mathematical model with the identified parameters. / Master of Science

This research investigates signal processing involving a single electromagnetic vector-sensor, with an emphasis on the problem regarding signal-selective narrowband direction-of-arrival (DOA) estimation in the presence of interference. The approach in this thesis relies on a high-resolution ESPRIT-based algorithm. Unlike spatially displaced arrays, the sensor cannot estimate the DOA of sources using phase differences between the array elements, as the elements are spatially co-located. However, the sensor measures the full electromagnetic field vectors, so the DOA can be estimated through the Poynting vector. Limited information is available in the open literature regarding signal-selective DOA estimation for a single electromagnetic vector-sensor. In this thesis, it is shown how the Uni-Vector-Sensor-ESPRIT (UVS-ESPRIT) algorithm that relies on a time-series invariance and was originally devised for deterministic harmonic sources can be applied to non-deterministic sources. Additionally, two algorithms, one based on cyclostationarity and the other based on fourth-order cumulants, are formulated based on the UVS-ESPRIT algorithm and are capable of selectively estimating the source DOA in the presence of interference based on the statistical properties of the sources. The cyclostationarity-based UVS-ESPRIT algorithm is capable of selectively estimating the signal-of-interest DOA when the sources have the same carrier frequency, and thus overlap in frequency. The cumulant-based UVS-ESPRIT algorithm devised for this sensor relies on the independent component analysis algorithm JADE and is capable of selectively estimating the signal-of-interest DOA through the fourth-order cumulants only, is robust to spatially colored noise, and is capable of estimating the DOA of more sources than sensor elements. / Master of Science / Electromagnetic vector-sensors are specialized sensors capable of capturing the full electromagnetic field vectors at a single point in space. Direction-of-arrival (DOA) estimation is the problem of estimating the spatial-angular parameters of one or more wavefronts impinging on an array. For a single electromagnetic vector-sensor, the array elements are not spatially displaced, but it is still possible to estimate the direction-of-arrival through the Poynting vector, which relates the electric and magnetic field vectors to the direction of propagation of an electromagnetic wave. Although direction-of-arrival estimation is a well-established area of research, there is limited discussion in the open literature regarding signal-selective DOA estimation in the presence of interference for a single electromagnetic vector-sensor. This research investigates this problem and discusses how the high-resolution Uni-Vector-Sensor-ESPRIT (UVS-ESPRIT) algorithm may be applied to non-deterministic sources. ESPRIT based algorithms capable of selectively estimating the source DOA are formulated based on the cyclostationarity and higher-order statistics of the sources, which are approaches known to be robust to interference. The approach based on higher-order statistics is also robust to spatially colored noise and is capable of estimating the DOA of more sources than sensor elements. The formulation of the UVS-ESPRIT for higher-order statistics relies on the application of the independent component analysis algorithm JADE, an unsupervised learning technique. Overall, this research investigates signal-selective direction-of-arrival estimation using an ESPRIT-based algorithm for a single electromagnetic vector-sensor.

"Blind" inversion of the effects of a given operator on a signal is an extremely difficult task that has no easy solutions. However,. Dr. Hany Farid has published several works that each individua:lly appear to achieve exactly this seemingly impossible result. In this work, we contribute a comprehensive overview of the published applications of blind process inversion, as well as provide the generalized form of the algorithms and requirements that are found in each of these applications, thereby formulating and explaining a general framework for blind process inversion using Farid's Algorithm. Additionally, we explain the knowledge required to derive the ROSA-based cost function on which Farid's Algorithm depends. As our primary contribution, we analyze the algorithmic complexity of this cost function based on the way it is currently, naively calculated, and derive a new algorithm to compute this cost function that has greatly reduced algorithmic complexity. Finally, we suggest an additional application of Farid's Algorithm to the problem of blindly estimating true camera response functions from a single image.

Neste trabalho, é apresentado e discutido um novo modelo para proteção de Linhas de Transmissão. O sistema proposto executa, individualmente, as etapas tradicionais da filosofia de proteção de distância: detecção, classificação e localização. Este modelo emprega Estatísticas de Ordem Superior (EOS) como ferramenta de extração de características, para posterior aplicação das Redes Neurais Artificiais (RNAs). As RNAs são responsáveis pelas tomadas de decisões do sistema, no sentido de identificar a ocorrência da falta e o tipo da mesma, além de localizar a falta no que tange às zonas de proteção consideradas. O processamento com tais estatísticas é responsável pela transformação dos dados para um domínio onde as diferentes faltas são evidenciadas através de agrupamentos de dados (padrões). O banco de dados disponível com sinais elétricos de LTs em condições de falta é utilizado para cálculo das estatísticas e o posterior treinamento supervisionado (e validação) das redes. A junção das etapas de proteção em um único modelo permitiu o desenvolvimento de um protótipo de relé, sendo executada uma bateria extensiva de testes, com as mais diversas condições de faltas possíveis. Apesar de operar apenas com sinais de corrente, o método proposto alcançou resultados que, em comparação com a técnica tradicional de proteção de distância, baseada na impedância aparente, aumenta consideravelmente o desempenho da proteção de LTs. Especialmente para as faltas monofásicas, de ocorrência mais comum, o desempenho obtido com o algoritmo proposto é largamente superior ao obtido com um relé de distância tradicional normalmente empregado em proteção de LTs, evidenciando a relevância da técnica empregada em aplicações de proteção. / A novel method of Transmission Lines (TLs) protection is presented and discussed in this work. The proposed algorithm performs the traditional steps of distance relaying, such as: fault detection, classification and location. The new method applies the Higher Order Statistics (HOS), also known as cumulants, as a tool for feature extraction in order to apply Artificial Neural Networks (ANN) for pattern classification. These networks are responsible for the processing of information, identifying a possible fault condition, the type of fault and, finally, its location in terms of fault zones considered for the problem. The application of HOS in a protection scheme is responsible for the transformation of electrical data, such as current signals, to a different domain where the different types of faults are highlighted by different classes of samples. The available database was obtained by simulating an Electric Power System and it is used for computing the statistics and training/validating the distinct neural networks of each step of the distance protection. A relay prototype is obtained by combining these steps in a synchronized operation. This prototype allowed the execution of extensive tests, simulating the operation of a protective system in real-time. Despite the use of currents signals only, the proposed method provided efficient protection for the EPS under study. In fact, comparing the results with a traditional method applied to distance protection, based on apparent impedance, an improvement of the protection performance was demonstrated. Especially for faults involving one phase and the ground, the most common in power systems, the results of the new methodology was significantly superior to that of the conventional relay. It can be concluded that the technique presents a high relevance for applications in transmission line protection.

Neste trabalho, é apresentado e discutido um novo modelo para proteção de Linhas de Transmissão. O sistema proposto executa, individualmente, as etapas tradicionais da filosofia de proteção de distância: detecção, classificação e localização. Este modelo emprega Estatísticas de Ordem Superior (EOS) como ferramenta de extração de características, para posterior aplicação das Redes Neurais Artificiais (RNAs). As RNAs são responsáveis pelas tomadas de decisões do sistema, no sentido de identificar a ocorrência da falta e o tipo da mesma, além de localizar a falta no que tange às zonas de proteção consideradas. O processamento com tais estatísticas é responsável pela transformação dos dados para um domínio onde as diferentes faltas são evidenciadas através de agrupamentos de dados (padrões). O banco de dados disponível com sinais elétricos de LTs em condições de falta é utilizado para cálculo das estatísticas e o posterior treinamento supervisionado (e validação) das redes. A junção das etapas de proteção em um único modelo permitiu o desenvolvimento de um protótipo de relé, sendo executada uma bateria extensiva de testes, com as mais diversas condições de faltas possíveis. Apesar de operar apenas com sinais de corrente, o método proposto alcançou resultados que, em comparação com a técnica tradicional de proteção de distância, baseada na impedância aparente, aumenta consideravelmente o desempenho da proteção de LTs. Especialmente para as faltas monofásicas, de ocorrência mais comum, o desempenho obtido com o algoritmo proposto é largamente superior ao obtido com um relé de distância tradicional normalmente empregado em proteção de LTs, evidenciando a relevância da técnica empregada em aplicações de proteção. / A novel method of Transmission Lines (TLs) protection is presented and discussed in this work. The proposed algorithm performs the traditional steps of distance relaying, such as: fault detection, classification and location. The new method applies the Higher Order Statistics (HOS), also known as cumulants, as a tool for feature extraction in order to apply Artificial Neural Networks (ANN) for pattern classification. These networks are responsible for the processing of information, identifying a possible fault condition, the type of fault and, finally, its location in terms of fault zones considered for the problem. The application of HOS in a protection scheme is responsible for the transformation of electrical data, such as current signals, to a different domain where the different types of faults are highlighted by different classes of samples. The available database was obtained by simulating an Electric Power System and it is used for computing the statistics and training/validating the distinct neural networks of each step of the distance protection. A relay prototype is obtained by combining these steps in a synchronized operation. This prototype allowed the execution of extensive tests, simulating the operation of a protective system in real-time. Despite the use of currents signals only, the proposed method provided efficient protection for the EPS under study. In fact, comparing the results with a traditional method applied to distance protection, based on apparent impedance, an improvement of the protection performance was demonstrated. Especially for faults involving one phase and the ground, the most common in power systems, the results of the new methodology was significantly superior to that of the conventional relay. It can be concluded that the technique presents a high relevance for applications in transmission line protection.

La maladie veineuse thromboembolique (MVTE) est un problème de santé publique (plus de 100000 cas par an en France). Elle regroupe deux entités cliniques : la thrombose veineuse profonde (TVP) des membres inférieurs et l’embolie pulmonaire (EP). La TVP correspond à la formation inadaptée d’un thrombus veineux (appelé aussi caillot sanguin) dans les veines profondes (poplitées, fémorales, iliaques). Un thrombus est principalement constitué de globules rouges et de plaquettes dans un réseau de fibrine. La complication majeure d’une TVP est la survenue d’une EP, c’est-à-dire que le thrombus s’est détaché de la paroi veineuse, ou s’est fragmenté, et est entraîné par la circulation sanguine jusqu’à une artère pulmonaire. Cette complication a un taux de mortalité assez élevée autour de 10000 à 20000 cas mortels par an en France. La survenue d’une TVP est multifactorielle associant des facteurs génétiques et acquis pouvant être répartis en trois catégories : la stase veineuse, l’altération de la paroi d’une veine et une hypercoagulabilité. En analysant la structure du thrombus, notre projet vise à identifier le facteur principal responsable de la TVP et à évaluer le risque d’EP. Pour caractériser sa structure, nous disposons de deux modes d’imagerie acoustique : l’échographie et l’élastographie (carte de dureté). Nous proposons d’extraire des descripteurs de ces images acoustiques par deux approches, l’une basée sur les ondelettes (le scattering operator) et l’autre sur les statistiques d’ordre supérieur (les multicorrélations). Ces descripteurs sont ensuite analysées par diverses techniques de classification (analyse en composantes principales, k-moyennes, classification spectrale) pour retrouver la cause principale des TVP ou la présence d’EP. / Venous thromboembolism (VTE) is an important public health issue (over 100000 individuals in France per year). VTE is a combination of a deep venous thrombosis (DVT) and a pulmonary embolism (PE). DVT is an inappropriate formation of a thrombus (also called blood clot) in one of the deep veins of the body, usually in the leg (popliteal, femoral, iliac). There are mainly three components in a thrombus: platelets, red blood cells and a mesh of fibrins. The main complication of a DVT is a pulmonary embolism (PE) which occurs when a thrombus breaks loose and travels to the lungs. PE affects an estimated 10000 - 20 000 individuals just in France per year. Three physiopathological mechanisms cancontribute, isolated or combined, to the development of a DVT: venous stasis, endothelial injury and hypercoagulability. Our project is aiming to relate the thrombus structure, its main triggering factor and the risk of a PE. To characterize the thrombus structure, we are collecting ultrasonography (echogenicity) and elastography (stiffness) of human thrombus. We propose to extract features from these to kind of ultrasound images using two approaches: one basedon wavelets (the scattering operator) and another based on high order statistics (multicorrelations). Then, the obtained features are analysed using several classification technics (principal component analysis, k-means, spectral clustering) to find the main cause of the DVT or the presence of PE.

Generating a single order statistic without generating the full sample can be an important task for simulations. If the density and the CDF of the distribution are given it is no problem to compute the density of the order statistic. In the main theorem it is shown that the concavity properties of that density depend directly on the distribution itself. Especially for log-concave distributions all order statistics have log-concave distributions themselves. So recently suggested automatic transformed density rejection algorithms can be used to generate single order statistics. This idea leads to very fast generators. For example for the normal and gamma distribution the suggested new algorithms are between 10 and 60 times faster than the algorithms suggested in the literature. (author's abstract) / Series: Preprint Series / Department of Applied Statistics and Data Processing

In this thesis, we proposed a precoding scheme for semi-blind channel estimation in amplify-and-forward (AF) multipath two-way relay networks (TWRN), where two terminals exchange their information through multi-relays. The precoding scheme, which diminishes computational complexity of semi-blind channel estimator, is used to distinguish received signal at both terminals from multi-relays. By applying a non-redundant linear precoding scheme at multi-relays, we proposed a semi-blind channel estimation to estimate the channel impulse response (CIR) of direct link and the cascaded source-relay-terminal links. Firstly, semi-blind channel estimation adopts least-square (LS) estimation to find the CIR of direct link between both terminals using a smaller number of training symbols. Secondly, the CIR of the cascaded source-relay-terminal links are obtained through second-order statistics (SOS) of received signals at both terminals. Consequently, the proposed scheme can effectively reduce the computational complexity and enhance the spectral efficiency in overall system. Simulation results corroborate the effectiveness of the proposed scheme.