Dissertations / Theses on the topic 'Distributed estimate'

Due to recent changes in agricultural practices the nutrient load in our waterways has increased causing eutrophication and hypoxia. Many legislative actions have taken place because of this problem, including the Clean Water Act of 1972 (CWA), and many different nutrient reduction plans. The CWA governs that impaired waterways must be monitored to meet total maximum daily loads (TMDL) for each watershed. TMDL's must be assessed using data collected over a period of time so that reduction techniques can be administered. TMDL assessments are usually conducted by the United States Geological Survey (USGS) through many different monitoring programs. The USGS programs include collecting streamflow and nutrient concentration data and using the information to estimate nutrient loads. Generally, grab sampling is the method of choice for concentrations. Grab samples do not accurately assess the total load as generally only 6-8 samples can be collected over a year due to financial and logistical constraints, while concentrations vary within a span of hours and days. Research applications involve the use of automated sensors (e.g., ISCO) that allow for more frequent sampling in order to overcome this issue but are expensive to purchase and maintain. Thus the development of an inexpensive, passive sampler would be of much interest in estimating load. The Passive Surface Flux Meter (PSFM), an integrative sampler that estimates the total solute load over a storm event, is such an alternative. The PSFM is composed of two sorbents one to collect the contaminant of choice and another to determine the flow through the device. Ion-exchange resin was used to collect nitrates, while Granular Activated Carbon dosed with a suite of alcohols were used to determine flow through the sampler. This thesis sets forth the fundamental theories behind the PSFM, and investigates its ability to measure nutrient fluxes in the field. In-situ deployments within Clear Creek watershed in Iowa were conducted using a modification of the PSFM design by Boland (2011). There was a strong linear relationship between the loads estimated by the PSFM, and "true" load based on USGS stream gage data, and Nitratax sensor data. This implies that the device could be calibrated to work in the field. However, it was determined that the design underestimated the true load in the stream by 29%. This was attributed to the nonlinear relationship between the external velocity and the flow through the sampler, which weighted the results towards the high flow events. To overcome this constraint, a new design is proposed in which flow through the sampler varies linearly with the transient head at the inlet. Flume experiments done under different flow depths proved that linearity conditions were satisfied. Using the results from the laboratory experiments recommendations were made for design of an in-situ deployment of the new design.

Recent advances in satellite sensor technology and micrometeorological instrumentation for water flux measurement, coupled with the expansion of automatic weather station networks that provide routine measurements of near-surface climate variables, present new opportunities for combining satellite and ground-based instrumentation to obtain distributed estimates of vegetation water use over wide areas in South Africa. In this study, a novel approach is tested, which uses satellite leaf area index (LAI) data retrieved by the Moderate Resolution Imaging Spectroradiometer (MODIS) to inform the FAO-56 Penman-Monteith equation for calculating reference evaporation (ET₀) of vegetation phenological activity. The model (ETMODIS) was validated at four sites in three different ecosystems across the country, including semi-arid savanna near Skukuza, mixed community grassland at Bellevue, near Pietermaritzburg, and Groenkop, a mixed evergreen indigenous forest near George, to determine potential for application over wider areas of the South African land surface towards meeting water resource management objectives. At Skukuza, evaluated against 170 days of flux data measured at a permanent eddy covariance (EC) flux tower in 2007, the model (ETMODIS) predicted 194.8 mm evapotranspiration relative to 148.9 mm measured fluxes, an overestimate of 31.7 %, (r² = 0.67). At an adjacent site, evaluated against flux data measured on two discrete periods of seven and eight days in February and May of 2005 using a large aperture scintillometer (SLS), ETMODIS predicted 27.4 mm and 6.7 mm evapotranspiration respectively, relative to measured fluxes of 32.5 and 8.2 mm, underestimates of 15.7 % and 18.3 % in each case (r² = 0.67 and 0.34, respectively). At Bellevue, evaluated against 235 days of evapotranspiration data measured using a surface layer scintillometer (SLS) in 2003, ETMODIS predicted 266.9 mm evapotranspiration relative to 460.2 mm measured fluxes, an underestimate of 42 % (r² = 0.67). At Groenkop, evaluated against data measured using a SLS over three discrete periods of four, seven and seven days in February, June and September/October respectively, ETMODIS predicted 9.7 mm, 10.3 mm and 17.0 mm evapotranspiration, relative to measured fluxes of 10.9 mm, 14.6 mm and 23. 9 mm, underestimates of 22.4 %, 11.2 % and 24.1 % in each case (r² = 0.98, 0.43 and 0.80, respectively). Total measured evapotranspiration exceeded total modelled evapotranspiration in all cases, with the exception of the flux tower site at Skukuza, where evapotranspiration was overestimated by ETMODIS by 31.7 % relative to measured (EC) values for the 170 days in 2007 where corresponding modelled and measured data were available. The most significant differences in measured versus predicted data were recorded at the Skukuza flux tower site in 2007 (31.7 % overestimate), and the Bellevue SLS flux site in 2003 (42 % underestimate); coefficients of determination, a measure of the extent to which modelled data are able to explain observed data at validation periods, with just two exceptions, were within a range of 0.67 – 0.98. Several sources of error and uncertainty were identified, relating predominantly to uncertainties in measured flux data used to evaluate ETMODIS, uncertainties in MODIS LAI submitted to ETMODIS, and uncertainties in ETMODIS itself, including model assumptions, and specific uncertainties relating to various inputs; further application of the model is required to test these uncertainties however, and establish confidence limits in performance. Nevertheless, the results of this study suggest that the technique is generally able to produce estimates of vegetation water use to within reasonably close approximations of measurements acquired using micrometeorological instruments, with r² values within the range of other peer-reviewed satellite remote sensing-based approaches.

Modern applications of mobile robots require them to have the ability to safely and effectively navigate in human environments. New challenges arise when these robots must plan their motion in a human-aware fashion. Current methods addressing this problem have focused mainly on the activity forecasting aspect, aiming at improving predictions without considering the active nature of the interaction, i.e. the robot’s effect on the environment and consequent issues such as reciprocity. Furthermore, many methods rely on computationally expensive offline training of predictive models that may not be well suited to rapidly evolving dynamic environments. This thesis presents a novel approach for enabling autonomous robots to navigate socially in environments with humans. Following formulations of the inverse planning problem, agents reason about the intentions of other agents and make predictions about their future interactive motion. A technique is proposed to implement counterfactual reasoning over a parametrised set of light-weight reciprocal motion models, thus making it more tractable to maintain beliefs over the future trajectories of other agents towards plausible goals. The speed of inference and the effectiveness of the algorithms is demonstrated via physical robot experiments, where computationally constrained robots navigate amongst humans in a distributed multi-sensor setup, able to infer other agents’ intentions as fast as 100ms after the first observation. While intention inference is a key aspect of successful human-robot interaction, executing any task requires planning that takes into account the predicted goals and trajectories of other agents, e.g., pedestrians. It is well known that robots demonstrate unwanted behaviours, such as freezing or becoming sluggishly responsive, when placed in dynamic and cluttered environments, due to the way in which safety margins according to simple heuristics end up covering the entire feasible space of motion. The presented approach makes more refined predictions about future movement, which enables robots to find collision-free paths quickly and efficiently. This thesis describes a novel technique for generating "interactive costmaps", a representation of the planner’s costs and rewards across time and space, providing an autonomous robot with the information required to navigate socially given the estimate of other agents’ intentions. This multi-layered costmap deters the robot from obstructing while encouraging social navigation respectful of other agents’ activity. Results show that this approach minimises collisions and near-collisions, minimises travel times for agents, and importantly offers the same computational cost as the most common costmap alternatives for navigation. A key part of the practical deployment of such technologies is their ease of implementation and configuration. Since every use case and environment is different and distinct, the presented methods use online adaptation to learn parameters of the navigating agents during runtime. Furthermore, this thesis includes a novel technique for allocating tasks in distributed robotics systems, where a tool is provided to maximise the performance on any distributed setup by automatic parameter tuning. All of these methods are implemented in ROS and distributed as open-source. The ultimate aim is to provide an accessible and efficient framework that may be seamlessly deployed on modern robots, enabling widespread use of intention prediction for interactive navigation in distributed robotic systems.

The spatial patterns of land surface and subsurface characteristics determine the spatial heterogeneity of hydrological processes. Soil depth is one of these characteristics and an important input parameter required by distributed hydrological models that explicitly represent spatial heterogeneity. Soil is related to topography and land cover due to the role played by topography and vegetation in affecting soil-forming processes. The research described in this dissertation addressed the development of statistical models that predict the soil depth pattern over the landscape; derivation of new topographic variables evaluated using both serial and parallel algorithms; and evaluation of the impacts of detailed soil depth representation on simulations of stream flow and soil moisture. The dissertation is comprised of three papers. In paper 1, statistical models were developed to predict soil depth pattern over the watershed based on topographic and land cover variables. Soil depth was surveyed at locations selected to represent the topographic and land cover variation at the Dry Creek Experimental Watershed, near Boise, Idaho. Explanatory variables were derived from a digital elevation model and remote sensing imagery for regression to the field data. Generalized Additive and Random Forests models were developed to predict soil depth over the watershed. The models were able to explain about 50% of the soil depth spatial variation, which is an important improvement over the soil depth extracted from the SSURGO national soil database. In paper 2, definitions of the new topographic variables derived in the effort to model soil depth, and serial and Message Passing Interface parallel implementations of the algorithms for their evaluation are presented. The parallel algorithms enhanced the processing speed of large digital elevation models as compared to the serial recursive algorithms initially developed. In paper 3, the impact of spatially explicit soil depth information on simulations of stream flow and soil moisture as compared to soil depth derived from the SSURGO soil database has been evaluated. The Distributed Hydrology Vegetation Soil Model was applied using automated parameter optimization technique with all input parameters the same except soil depth. Stream flow was less impacted by the detailed soil depth information, while simulation of soil moisture was slightly improved due to the detailed representation of soil depth.

付記する学位プログラム名: グローバル生存学大学院連携プログラム
Kyoto University (京都大学)
0048
新制・課程博士
博士(工学)
甲第22771号
工博第4770号
新制||工||1746(附属図書館)
京都大学大学院工学研究科機械理工学専攻
(主査)教授 松野 文俊, 教授 椹木 哲夫, 教授 泉田 啓
学位規則第4条第1項該当

Submitted by Maria Aparecida (cidazen@gmail.com) on 2017-04-17T13:14:17Z No. of bitstreams: 1 Elson Moreira.pdf: 7883984 bytes, checksum: cf59b3b0b24a249a7fd9e2390b7f16de (MD5)
Made available in DSpace on 2017-04-17T13:14:17Z (GMT). No. of bitstreams: 1 Elson Moreira.pdf: 7883984 bytes, checksum: cf59b3b0b24a249a7fd9e2390b7f16de (MD5) Previous issue date: 2017-02-10
CNPQ
A problem of electric power quality that always affects the consumers of the distribution network are the harmonic distortions. Harmonic distortions arise from the presence of socalled harmonic sources, which are nonlinear equipment, i.e., equipment in which the voltage waveform differs from the current. Such equipment injects harmonic currents in the network generating distortions in the voltage waveform. Nowadays, the number of these equipment in the electrical network has increased considerably. However, the increasing use of such equipment over the network makes systems more vulnerable and prone to quality problems in the supply of electricity to consumers. In addition, it is important to note that in the current scenario, the generation of electricity from renewable sources, connected in the secondary distribution network, is increasing rapidly. This is mainly due to shortage and high costs of fossil fuels. In this context, the Photovoltaic Distributed Generation (PVDG), that uses the sun as a primary source for electric energy generation, is the main technology of renewable generation installed in distribution network. However, the PVDG is a potential source of harmonics, because the interface of the PVDG with the CA network is carried out by a CC/CA inverter, that is a highly nonlinear equipment. Thus, the electrical power quality problems associated with harmonic distortion in distribution networks tend to increase and be very frequent. One of the main indicators of harmonic distortion is the total harmonic distortion of voltage ( ) used by distribution utilities to limit the levels of harmonic distortion present in the electrical network. In the literature there are several deterministic techniques to estimate . These techniques have the disadvantage of not considering the uncertainties present in the electric network, such as: change in the network configuration, load variation, intermittence of the power injected by renewable distributed generation. Therefore, in order to provide a more accurate assessment of the harmonic distortions, this dissertation has as main objective to develop a probabilistic methodology to estimate the level of in secondary distribution networks considering the uncertainties present in the network and PVDG connected along the network. The methodology proposed in this dissertation is based on the combination of the following techniques: three-phase harmonic power flow in phase coordinate via method sum of admittance, point estimate method and series expansion of Gram-Charlier. The validation of the methodology was performed using the Monte Carlo Simulation. The methodology was tested in European secondary distribution network with 906 nodes of 416 V. The results were obtained by performing two case studies: without the presence of PVDG and with the PVDG connection. For the case studies, the following statistics for nodal were estimated: mean value, standard deviation and the 95% percentile. The results showed that the probabilistic estimation of is more complete, since it shows the variation of due to the uncertainties associated with harmonic sources and electric network. In addition, they show that the connection of PV-DG in the electric network significantly affects the levels of of the electric network.
Um problema de qualidade de energia elétrica que afeta os consumidores da rede de distribuição secundária são as distorções harmônicas. As distorções harmônicas são provenientes da presença das chamadas fontes de harmônicas que são equipamentos de características não-lineares, ou seja, equipamentos em que a forma de onda da tensão difere da de corrente. Tais equipamentos injetam correntes harmônicas na rede produzindo, portanto distorções na forma de onda da tensão. Nos dias atuais, a quantidade desses equipamentos na rede elétrica tem aumentado consideravelmente. Porém, o uso crescente desse tipo de equipamento ao longo da rede torna os sistemas mais vulneráveis e propensos a apresentarem problemas de qualidade no fornecimento de energia elétrica aos consumidores. Além disso, é importante destacar que no cenário atual, a geração de energia elétrica a partir de fontes renováveis, conectada na rede de distribuição secundária, está aumentando rapidamente. Isso se deve principalmente devido a escassez e altos custos dos combustíveis fosseis. Neste contexto, a Geração Distribuída Fotovoltaica (GDFV), que utiliza o sol como fonte primária para geração de energia elétrica, é a principal tecnologia de geração renovável instalada na rede de distribuição no Brasil. Contudo, a GDFV é uma potencial fonte de harmônica, pois a interface da GDFV com a rede CA é realizada por um inversor CC/CA, que é um equipamento altamente não-linear. Desde modo, os problemas de qualidade de energia elétrica associados à distorção harmônica nas redes de distribuição tendem a aumentar e a serem bem frequentes nos consumidores da rede de distribuição secundárias. Um dos principais indicadores de distorção harmônica é a distorção harmônica total de tensão ( do inglês “Total Harmonic Distortion of Voltage”) utilizada pelas concessionárias de energia elétrica para quantificar os níveis de distorção harmônica presentes na rede elétrica. Na literatura técnica existem várias técnicas determinísticas para estimar a . Essas técnicas possuem a desvantagem de não considerar as incertezas presentes na rede elétrica, tais como: mudança na configuração da rede, variação de carga e intermitência da potência injetada pela geração distribuída renovável. Portanto, a fim de fornecer uma avaliação mais precisa das distorções harmônicas, este trabalho tem como principal objetivo desenvolver uma metodologia probabilística para estimar o nível de em redes de distribuição secundária considerando as incertezas presentes na rede e na GDFV conectada ao longo da rede. A metodologia proposta nesta dissertação se baseia na combinação das seguintes técnicas: fluxo de potência harmônico trifásico em coordenadas de fase via método de soma de admitância, método de estimação por pontos e expansão em série de Gram-Charlier. Além disso, a validação da metodologia foi realizada utilizando a Simulação Monte Carlo. A metodologia desenvolvida foi testada na rede de distribuição secundária europeia com 906 nós de 416 V. Os resultados foram obtidos realizando dois casos de estudos: sem a presença de GDFV e com a conexão de GDFV. Para ambos os casos de estudo as seguintes estatísticas do nodal foram estimadas: valor médio, desvio padrão e o percentil de 95%. Os resultados demonstraram que a estimação probabilística da é mais completa, pois mostra a variação da devido às incertezas associadas com as fontes de harmônicas e as da rede elétrica. Os resultados também mostram que a conexão da GDFV afeta significativamente os níveis de da rede elétrica

Recently, a lot of research work has been dedicated toward enhancing performance, reliability and integrity of distributed energy resources that are integrated into distribution networks. The problem of islanding detection and islanding prevention (i.e. anti-islanding) has stimulated a lot of research due to its role in severely compromising the safety of working personnel and resulting in equipment damages. Various Islanding Detection Methods (IDMs) have been developed within the last ten years in anticipation of the tremendous increase in the penetration of Distributed Generation (DG) in distribution system. This work proposes new IDMs that rely on transient and distributed behaviors to improve integrity and performance of DGs while maintaining multi-DG islanding detection capability. In this thesis, the following questions have been addressed: How to utilize the transient behavior arising from an islanding condition to improve detectability and robust performance of IDMs in a distributive manner? How to reduce the negative stability impact of the well-known Sandia Frequency Shift (SFS) IDM while maintaining its islanding detection capability? How to incorporate the perturbations provided by each of DGs in such a way that the negative interference of different IDMs is minimized without the need of any type of communication among the different DGs? It is shown that the proposed techniques are local, scalable and robust against different loading conditions and topology changes. Also, the proposed techniques can successfully distinguish an islanding condition from other disturbances that may occur in power system networks. This work improves the efficiency, reliability and safety of integrated DGs, which presents a necessary advance toward making electric power grids a smart grid.
Ph.D.
Doctorate
Electrical Engineering and Computer Science
Engineering and Computer Science
Electrical Engineering

The issue of forecasting the future price of risky financial assets has attracted academia and business practice since the inception of the stock exchange. Also due to the just finished financial crisis, which was the worst crisis since the Great Depression, it is clear that research in this area has not been finished yet. On the contrary, new challenges have been raised. The main goal of the thesis is the demonstration of the significant information potential which is hidden in option market prices. These prices contain informations on probability distribution of the underlying asset returns and the risk connected with these returns. Other objectives of the thesis are the forecast of the underlying asset price distribution using parametric and nonparametric estimates, the improvement of this forecast using the utility function of the representative investor, the description of the current market sentiment and the determination of the risk premium, especially the risk premium on Czech market. The thesis deals with the forecast of the underlying asset price probability distribution implied by the current option market prices using parametric and nonparametric estimates. The resulting distribution is described by the moment characteristics which represent a valuable tool for analyzing the current market sentiment. According to the theory, the probability distribution of the underlying asset price implied by option prices is risk neutral, i.e. it applies only to risk neutral investors. The theory further implies that the distribution of real world can be derived from the risk neutral distribution using utility function of the representative investor. The inclusion of a utility function of representative investor improves the forecast of the underlying asset price distribution. Three different utility functions of traditional risk theory are used in the thesis. These functions range from the simple power function to the general function of hyperbolic absolute risk aversion (HARA). Further, Friedman-Savage utility function is used. This function allows both a risk averse investor and a risk loving investor. The thesis also answers the question: Are the current asset prices at so high level that the purchase of the asset means a gamble? The risk premium associated with investing in the risky asset is derived in the thesis. The risk premium can be understood as the premium demanded by investors for investment in a risky asset against the investment in a riskless asset. All the theoretical methods introduced in the thesis are demonstrated on real data coming from two different markets. Developing market is represented by shares of CEZ and developed market is represented by S&P 500 futures. The thesis deals with demonstrations in single point in time as well as in available history of the data. The forecasts of the underlying asset price distribution and the relating risk premium are constructed in the available data history. The goals and the objectives of the thesis have been achieved. The contribution of the thesis is the development of parametric and nonparametric methodology for estimating the underlying asset price probability distribution implied by the option market prices so that the nature of the particular market and instrument is captured. The further contribution of the thesis is the construction of the forecasts of the underlying asset price distribution and the construction of the market sentiment in the available history of data. The contribution of the thesis is also the construction of the market risk premium in the available history and the establishment of the hypothesis that the markets gamble before the crisis.

This dissertation develops optimal algorithms for distributed detection and estimation in static and mobile sensor networks. In distributed detection or estimation scenarios in clustered wireless sensor networks, sensor motes observe their local environment, make decisions or quantize these observations into local estimates of finite length, and send/relay them to a Cluster-Head (CH). For event detection tasks that are subject to both measurement errors and communication errors, we develop an algorithm that combines a Maximum a Posteriori (MAP) approach for local and global decisions with low-complexity channel codes and processing algorithms. For event estimation tasks that are subject to measurement errors, quantization errors and communication errors, we develop an algorithm that uses dithered quantization and channel compensation to ensure that each mote's local estimate received by the CH is unbiased and then lets the CH fuse these estimates into a global one using a Best Linear Unbiased Estimator (BLUE). We then determine both the minimum energy required for the network to produce an estimate with a prescribed error variance and show how this energy must be allocated amongst the motes in the network. In mobile wireless sensor networks, the mobility model governing each node will affect the detection accuracy at the CH and the energy consumption to achieve this level of accuracy. Correlated Random Walks (CRWs) have been proposed as mobility models that accounts for time dependency, geographical restrictions and nonzero drift. Hence, the solution to the continuous-time, 1-D, finite state space CRW is provided and its statistical behavior is studied both analytically and numerically. The impact of the motion of sensor on the network's performance is also studied.

Over the last couple of decades changes to pediatric inflammatory bowel disease (IBD) care may have altered health services use among these children. I used a retrospective matched cohort design and population-based health administrative data to first quantify trends in IBD health services and surgical outcomes in Ontario IBD children diagnosed between 1994-2012. I then used these results to validate the distributed network analysis method – a method being increasingly used in Canadian multi-province studies where privacy regulations prevent sharing of individual-level data across provincial borders - using Ontario’s Local Health Integration Networks. I found (1) decreasing hospitalizations and surgical outcomes but increasing outpatient visit rates, suggesting changing patterns of health care use in Ontario children with IBD, and, (2) distributed network analyses is a satisfactory privacy-preserving alternative to individual-level analysis under the conditions tested in my study, providing a tested analysis method for researchers using multi-jurisdictional data.

Dans cette thèse on s'intéresse à la contrôlabilité de deux classes de systèmes paraboliques linéaires.On caractérise dans un premier temps la contrôlabilité à zéro de systèmes à coefficients constants en dimension 1 lorsque les contrôles agissent sur différentes parties du domaine ou de sa frontière.On regarde ensuite avec le théorème de Fattorini la contrôlabilité frontière approchée de ces systèmes en dimension quelconque.On obtient notamment que les systèmes de 2 équations sont toujours contrôlables dans un rectangle si la zone de contrôle contient 2 directions.Dans un autre travail sur les systèmes à coefficients constants, on obtient une estimation du coût du contrôle frontière à zéro en dimension 1.On utilise ce résultat pour montrer que la contrôlabilité frontière à zéro dans des domaines cylindrique est réduite à la contrôlabilité frontière à zéro en dimension 1.On étudie ensuite la contrôlabilité approchée de systèmes en cascade avec un couplage d'ordre 1.On prouve que la contrôlabilité interne avec un couplage constant à toujours lieu, quel que soit la dimension et la zone de contrôle.On établit d'autre part une caractérisation de la contrôlabilité frontière en dimension 1 avec un couplage variable.Enfin, dans une dernière partie on s'intéresse à la contrôlabilité interne approchée de systèmes en cascade à coefficients variables en dimension 1.On montre qu'on est ramené à établir une caractérisation de la propriété de continuation unique pour une équation elliptique non-homogène.A l'aide de la caractérisation alors obtenue on montre en particulier comment la géométrie de la zone de contrôle peut influencer la contrôlabilité des systèmes
This thesis focuses on the controllability of two classes of linear parabolic systems.We start with a caracterization of the null-controllability of systems with constant coefficients in dimension 1 where the controls are acting on different parts of the domain or its boundary.With the help of the theorem of Fattorini we then look at the boundary approximate controllability of these systems in any dimension.We show that a system of 2 equations is always approximately controllable on a rectangle if we assume that the control domain contains 2 directions.In another work on the systems with constant coefficients, we obtain an estimate of the boundary null-control cost in dimension 1.We then use this result to show that the boundary null-controllability in cylindrical domains is reduced to the boundary null-controllability in dimension 1.We then study the approximate controllability of cascade systems with a first order coupling term.We prove the distributed controllability when the coupling is constant, whatever the dimension and control domain are.On the other hand, we establish a caracterisation of the boundary controllability in dimension 1 for space-dependent couplings.Last, we investigate the distributed approximate controllability of cascade systems with space-dependent coefficients in dimension 1.Using the theorem of Fattorini and the structure of the systems under study we are lead to characterize the unique continuation property for a non-homogeneous elliptic equation.With the help of the caracterization then obtained we show in particular how the geometry of the control domain can affect the controllability properties of systems

Nous analysons dans cette thèse différents aspects associés à la modélisation des écoulements à surface libre en eaux peu profondes (Shallow Water). Nous étudions tout d’abord le système d’équations de Saint-Venant à deux dimensions et leur résolution par la méthode numérique des volumes finis, en portant une attention particulière sur les aspects hyperboliques et conservatifs. Ces schémas permettent de traiter les équilibres stationnaires, les interfaces sec/mouillé et aussi de modéliser des écoulements subcritique, transcritique et supercritique. Nous présentons ensuite la théorie de la méthode d’assimilation variationnelle de données adaptée à ce type d’écoulement. Son application au travers des études de sensibilité est longuement discutée dans le cadre de l'hydraulique à surface libre. Après cette partie à caractère théorique, la partie tests commence par une qualification de l’ensemble des méthodes numériques qui sont implémentées dans le code DassFlow, développé à l’Université de Toulouse, principalement à l’IMT mais aussi à l’IMFT. Ce code résout les équations Shallow Water par une méthode de volumes finis et est validé par comparaison avec les solutions analytiques pour des cas tests classiques. Ces mêmes résultats sont comparés avec un autre code d’hydraulique à surface libre aux éléments finis en deux dimensions, Telemac 2D. Une particularité notable du code DassFlow est de permettre l’assimilation variationnelle de données grâce au code adjoint permettant le calcul du gradient de la fonction coût. Ce code adjoint a été obtenu en utilisant l'outil de différentiation automatique Tapenade (Inria). Nous testons ensuite sur un cas réel, hydrauliquement complexe, différentes qualités de Modèles Numériques de Terrain (MNT) et de bathymétrie du lit d’une rivière. Ces informations proviennent soit d’une base de données classique type IGN, soit d’informations LIDAR à très haute résolution. La comparaison des influences respectives de la bathymétrie, du maillage et du type de code utilisé, sur la dynamique d’inondation est menée très finement. Enfin nous réalisons des études cartographiques de sensibilité aux paramètres du modèle sur DassFlow. Ces cartes montrent l’influence respective des différents paramètres ou de la localisation des points de mesure virtuels. Cette localisation optimale de ces points est nécessaire pour une future assimilation de données efficiente
We analyze in this thesis various aspects associated with the modeling of free surface flows in shallow water approximation. We first study the system of Saint-Venant equations in two dimensions and its resolution with the numerical finite volumes method, focusing in particular on aspects hyperbolic and conservative. These schemes can process stationary equilibria, wetdry interfaces and model subcritical, transcritical and supercritical flows. After, we present the variational data assimilation method theory fitted to this kind of flow. Its application through sensitivity studies is fully discussed in the context of free surface water. After this theoretical part, we test the qualification of numerical methods implemented in the code Dassflow, developed at the University of Toulouse, mainly at l'IMT, but also at IMFT. This code solves the Shallow Water equations by finite volume method and is validated by comparison with analytical solutions for standard test cases. These results are compared with another hydraulic free surface flow code using finite elements in two dimensions: Telemac2D. A significant feature of the Dassflow code is to allow variational data assimilation using the adjoint method for calculating the cost function gradient. The adjoint code was obtained using the automatic differentiation tool Tapenade (INRIA). Then, the test is carried on a real hydraulically complex case using different qualities of Digital Elevation Models (DEM) and bathymetry of the river bed. This information are provided by either a conventional database types IGN or a very high resolution LIDAR information. The comparison of the respective influences of bathymetry, mesh size, kind of code used on the dynamics of flooding is very finely explored. Finally we perform sensitivity mapping studies on parameters of the Dassflow model. These maps show the respective influence of different parameters and of the location of virtual measurement points. This optimal location of these points is necessary for an efficient data assimilation in the future

碩士
國立臺灣科技大學
營建工程技術學系
81
In architectural structure analysis,the simplified frame model is usually adopted for analyzed.And the determination of the equivalent distributed vertical load is almost based on the yielding line theory of slab,that is,to use partition zone of yielding line as the equivalent distributed vertical load on beam. The research of this thesis is to disuss the accuracy of the general loading mode as the above mentioned. At first, by the structural analysis software SAP-90, using the frame model that included shell element to analysis indiated loading condition.When the calculations for all the usually used aspect ratio is done,take the analyzed shear and torsion on the frame regressing to shear and torsional fitting curve. When the polynomial shear curve is differentiated , we can get the equivalent distributed vertical load , as the same , when differentiated the polymial torsion curve , we can get the distributed torsion on beam.As the distributed load for all the aspect ratio is get ,from appropriate numerial method, we can determine all the load on the simplified frame model. When the loading is determinated from the above mentioned method , to verify the analysis accuray from this method is another another important task of this thesis. It is the final destination of this thesis,that, if we can get more accurately analysis than the traditional load method.

Attualmente le reti di sensori rappresentano una tecnologia efficace per la realizzazione di sistemi di monitoraggio scalabili, flessibili, a costi contenuti e tolleranti ai guasti. La sorveglianza di un processo, di un oggetto o di un fenomeno richiede la stima dello stato del processo d’interesse mediante l’utilizzo di modelli matematici e l’effettuazione di misure di grandezze direttamente o indirettamente legate al suo stato. I sistemi di monitoraggio tradizionali sono costituiti da un insieme di sensori distribuiti nella regione di sorveglianza connessi con un unico centro di elaborazione che provvede a elaborare i dati raccolti. Negli ultimi tempi si stanno sempre più diffondendo reti in cui ogni nodo (o agente) è dotato di capacità di elaborazione, comunicazione e, in alcuni casi, rilevamento. Una rete di questo tipo è detta distribuita o anche net-centrica, in quanto ogni agente è in grado di effettuare una stima dello stato del processo monitorato, o di una sottoparte di esso. Due diversi approcci per la stima distribuita di sistemi lineari, detti consenso sull’informazione a posteriori (CP: Consensus on Posteriors) e consenso sulla verosi- miglianza (CL: Consensus on Likelihoods), vengono combinati per dar luogo ad una nuova classe di filtri di consenso su verosimiglianza e informazione a priori (CLCP: Consensus on Likelihoods and Priors), che beneficiano dei vantaggi complementari degli approcci CL e CP. Vengono presentati nuovi risultati teorici, limitatamente ai sistemi lineari, sulla stabilità dei filtri CLCP sotto condizioni di osservabilità collettiva e connettività di rete. Infine viene fatta una valutazione comparativa delle prestazini di alcuni filtri basati sul consenso in un caso di studio relativo al tracciamento di un oggetto in movimento, sia con sensori lineari che non lineari. Una delle principali vulnerabilità dei sistemi di sorveglianza net-centrici è la loro debolezza rispetto agli effetti distruttivi di virus informatici. Si sono studiati i principali modelli per la diffusione dei virus e l’insorgenza di un focolaio epidemico in relazione ai principali parametri di rete. Un ampio insieme di simulazioni su scenari realistici di sorveglianza ha permesso di analizzare il modo in cui i parametri di rete influenzano la diffusione del virus e la fusione distribuita dell’informazione, fornendo indicazioni utili per preservare la funzionalità di sorveglianza della rete in presenza di guasti e/o attacchi informatici. Nelle reti di monitoraggio e sorveglianza non è detto che ciascun agente abbia una visione complessiva dello stato del sistema. Si è studiato il problema della stima distribuita nel caso in cui gli agenti abbiano una visione parziale del sistema complessivo facendo uno studio comparativo tra varie tecniche di fusione dei dati su scenari di studio sia lineari che non lineari.

碩士
國立彰化師範大學
工業教育與技術學系
98
This thesis presents the complete study of stability analysis and state estimators design. The system is focused on neutral neural networks with both interval discrete and distributed time-varying delays, where the time-varying delays are in a given range. In a stability analysis problem, the purpose is to develop globally robust delay-dependent stability for neutral uncertain neural networks with both discrete and distributed delays. The activation functions are supposed to be bounded and globally Lipschitz continuous. By using a Lyapunov function approach and linear matrix inequality (LMI) techniques, the stability criteria for the neutral uncertain neural networks with both discrete and distributed delays are established in the form of LMIs, which can be readily verified by using standard numerical software. In an estimator design problem, the estimation for neutral neural network with both discrete and distributed interval time-varying delays is investigated. By using the Lyapunov-Krasovskii method, a linear matrix inequality (LMI) approach is developed to construct sufficient conditions for the existence of admissible state estimators such that the error-state system is globally asymptotically stable. Then, we show that both the existence conditions and the explicit expression of the desired estimator can be characterized in terms of the solution to an LMI. Finally, some illustrative examples have been presented to demonstrate the effectiveness of the proposed approach.

One of the major applications of wireless sensor networks is to sense accurate and reliable data from the physical environment with or without a priori knowledge of data statistics. To extract accurate data from the physical environment, we investigate spatial data correlation among sensor nodes to develop data accuracy models. We propose three data accuracy models namely Estimated Data Accuracy (EDA) model, Cluster based Data Accuracy (CDA) model and Distributed Cluster based Data Accuracy (DCDA) model with a priori knowledge of data statistics. Due to the deployment of high density of sensor nodes, observed data are highly correlated among sensor nodes which form distributed clusters in space. We describe two clustering algorithms called Deterministic Distributed Clustering (DDC) algorithm and Spatial Data Correlation based Distributed Clustering (SDCDC) algorithm implemented under CDA model and DCDA model respectively. Moreover, due to data correlation in the network, it has redundancy in data collected by sensor nodes. Hence, it is not necessary for all sensor nodes to transmit their highly correlated data to the central node (sink node or cluster head node). Even an optimal set of sensor nodes are capable of measuring accurate data and transmitting the accurate, precise data to the central node. This reduces data redundancy, energy consumption and data transmission cost to increase the lifetime of sensor networks. Finally, we propose a fourth accuracy model called Adaptive Data Accuracy (ADA) model that doesn't require any a priori knowledge of data statistics. ADA model can sense continuous data stream at regular time intervals to estimate accurate data from the environment and select an optimal set of sensor nodes for data transmission to the network. Data transmission can be further reduced for these optimal sensor nodes by transmitting a subset of sensor data using a methodology called Spatio-Temporal Data Prediction (STDP) model under data reduction strategies. Furthermore, we implement data accuracy model when the network is under a threat of malicious attack.

A presente dissertação pretende investigar a relação entre o desenvolvimento do setor bancário e o desenvolvimento do mercado acionista no contexto do sistema financeiro Português, no período entre 1978 e 2017. O setor bancário e o mercado acionista são, frequentemente, vistos como rivais, com alguns teóricos sugerindo que um deles desempenha um papel dominante na promoção do crescimento económico. Recentemente, o debate sobre o paradigma de superioridade alterou-se para a noção de interação entre bancos e mercados no sistema financeiro. Estimamos quatro modelos lineares, que incluem uma proxy para o desenvolvimento do setor bancário (crédito ao setor privado) e uma proxy para o desenvolvimento do mercado acionista (capitalização de mercado acionista). Também foram incluídas duas variáveis macroeconómicas de controlo (agregado monetário M3 e taxa de juro nominal de curto prazo). Utilizando o estimador autorregressivo de desfasamentos distribuídos (ARDL), encontramos uma relação positiva de longo e curto prazo entre o desenvolvimento do setor bancário e o desenvolvimento do mercado acionista, o qual sugere que no curto prazo a evolução do setor bancário estimula a evolução do mercado acionista e no longo prazo a evolução do mercado acionista estimula a evolução do setor bancário. Em outras palavras, o setor bancário e o mercado acionista coevoluem entre si, em vez de competirem. Como o sistema financeiro Português assenta, claramente, no setor bancário, recomendamos que sejam feitos esforços em direção a um desenvolvimento mais rápido e eficiente do mercado acionista Português. Deste modo, o sistema financeiro Português poderá retirar melhor proveito desta relação de coevolução.
This paper aims to assess the relationship between bank development and stock market development in the Portuguese financial system context, from 1978 to 2017. Banks and stock market are often viewed as competitors, with some theorists suggesting that one plays a dominant role in foster economic growth. Recently, there has been a paradigm shift in the debate, from superiority to the interaction between banks and markets in a financial system. We estimate four linear models, which includes one proxy for the bank development (credit to private sector) and one proxy for the stock market development (stock market capitalization). We also included two macroeconomic control variables (monetary aggregate M3 and nominal short-term interest rate). Using autoregressive distributed lag (ARDL) estimator, we find a positive long-run and short-run relationship between bank development and stock market development, which suggests that in short-term bank evolution stimulates stock market evolution and in long-term stock market evolution stimulates bank evolution. In the other words, banks and markets coevolve with each other rather than compete. As the Portuguese financial system is clearly bank-based, we recommend that efforts should be made toward the efficiently and faster development of Portuguese stock market. Thus, Portuguese financial system can take better benefit from this coevolve relationship.