O estudo de inventário tem por objetivo estimar o potencial hidroelétrico de rios ou bacias, analisando várias alternativas propostas de partição de quedas, sendo que cada alternativa contém um conjunto de aproveitamentos hidroelétricos. Essas alternativas são então estudadas individualmente para definição da alternativa ótima, ou seja, a que tem melhor custo beneficio e ao mesmo tempo cause menos danos ambientais. Para essa análise necessitamos calcular a potência de cada aproveitamento específico, assim como a energia gerada, para isso então precisamos conhecer a vazão do rio em estudo, no local desses aproveitamentos. Como a vazão dos rios varia com o tempo, pois depende de variáveis como clima, geologia dos solos, desmatamento, entre outras, se recomenda usar nos cálculos séries longas de vazões médias com no mínimo 30 anos de dados, o problema é que em muitos casos não temos essas séries ou temos séries menores e incompletas, nesse caso então necessitamos estimar os valores ausentes e ruidosos utilizando os dados de estações fluviométricas próximas, para depois transportá-las para o aproveitamento em estudo, para isso utilizamos de técnicas estatísticas de correlação. A ideia nesse trabalho é de utilizarmos redes neurais artificiais ao invés das técnicas convencionais e comparar os resultados obtidos. / The inventory study aims to estimate the hydropower potential of rivers or basins, analyzing several alternative proposals for partition of falls, each of which contains a set of alternative hydroelectric developments. These alternatives are then individually analyzed to define the optimal alternative, namely that which has the best cost benefit while causing less environmental damage. For this analysis we need to calculate the power of each specific use, as well as the energy generated for that then we need to know the flow of the river under study, the location of these usages. As the river flow varies with time because it depends on variables such as climate, geology, soils, deforestation, among others, we recommend using the long series of calculations mean flow at least 30 years of data, the problem is that in many cases we do not have these series or have smaller and incomplete series, in this case then we need to estimate the missing values and noisy data using next gauged stations, and then transport them to use in the study, for this we use statistical correlation techniques. The idea is that we use work instead of the conventional Artificial Neural Network techniques and compare the results.

Este trabalho contempla os estudos realizados para elaboração de uma carta de eventos perigosos (hazard) de uma área de aproximadamente 45 km², no município de Ouro Preto/MG, a partir de dados geológicos e geotécnicos, gerados em trabalhos de mapeamento geotécnico, com o uso dos recursos de redes neurais artificiais e da probabilidade condicional. Os dados prévios foram tratados e um conjunto de 15 mapas e cartas elaborado, a saber: topográfico, de substrato rochoso, material inconsolidado, de uso e ocupação, de inventário dos movimentos de massa gravitacionais (escorregamentos translacionais, escorregamentos translacionais tipo de material), de declividade, de rumo da inclinação das encostas, das unidades geológico-geotécnicas, das seções típicas das unidades geológico-geotécnicas, da resistência ao cisalhamento, do contraste de permeabilidade e da superfície potencial de ruptura, associado a uma tabela com as características das unidades geológico-geotécnicas. Os modelos de redes neurais artificiais e probabilidade condicional foram desenvolvidos para o uso em MATLAB utilizando um conjunto de 11 mapas e cartas dentre os citados anteriormente. A análise dos dados prévios frente aos modelos foi desenvolvida no sentido de avaliar a sua qualidade e a sua adequação ao modelo proposto. Concluiu-se sobre a necessidade de dados específicos que nem sempre são gerados em trabalhos rotineiros, como: levantamento da atividade, velocidade, volume e data de ocorrência, entre outros para caracterização das feições de movimentos de massa gravitacionais; estudo detalhado dos parâmetros de resistência dos materiais e das descontinuidades presentes no substrato rochoso; dados de estações pluviométricas para estudos da intensidade e distribuição da chuva na região, entre outras informações. / This work shows the studies developed for elaboration of the landslide hazard map of the area of 45 km², approximately, in the municipality of Ouro Preto, in the state of Minas Gerais, Brazil, from data generated in geotechnical mapping, with the use of artificial neural networks and conditional probability methods. The previous data were processed and was elaborated a set of 15 maps and charts: topographic, lithologies, unconsolidated material, land uses, inventory (landslides, translational slides, translational inventory - type of geological material, slope, slope inclination direction, geological - geotechnical units, typical topographic profile of the geological and geotechnical units, the shear strength categories, hydraulic conductivity contrasts, potential failure surfaces and a table with characteristic of the geological and geotechnical units. The procedures of the artificial neural networks and conditional probability were developed for use in MATLAB using a set of 11 maps among the 15 elaborated. A analysis of the previous data prepared and the data necessary for models was developed to evaluate its suitability. The main conclusion is that the routine mapping and inventories do not consider important attributes, such as activity, movement rate, volume, landslide date and others aspects of the features; detailed study about shear strength of geological materials and discontinuities and rainfall data.

Orientadores: Takaaki Ohishi, Ricardo Menezes Salgado / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenharia Elétrica e de Computação / Made available in DSpace on 2018-08-23T18:34:18Z (GMT). No. of bitstreams: 1 FrancoJunior_EdgarFonseca_M.pdf: 7364704 bytes, checksum: be747ce93528de5661be3b2b3bf77cbc (MD5) Previous issue date: 2013 / Resumo: Em um sistema de energia elétrica em corrente alternada, a geração, a transmissão e o consumo de energia elétrica são divididos em potência ativa e reativa. O planejamento, a operação e análise destes sistemas são baseados em estimativas futuras do consumo de energia, e neste contexto são importantes os modelos de previsão de carga ativa e reativa. Nesta dissertação são testados modelos de previsão de curto prazo para carga ativa e reativa utilizando modelos de redes neurais artificiais. Em particular, são implementados e testados várias metodologias de seleção de entradas. A seleção de um subconjunto apropriado de variáveis para a inclusão em um sistema é um passo vital no desenvolvimento de qualquer modelo. Isto é particularmente importante nos modelos de previsão como redes neurais artificiais, pois o desempenho do modelo final é fortemente dependente das variáveis de entrada utilizadas. Esta dissertação desenvolveu um modelo que dá suporte à integração de algumas técnicas de seleção (informação mútua e informação mútua parcial) tendo o intuito de facilitar a utilização destas, assim como a sua comparação quando aplicada a determinados problemas de previsão. Para os experimentos, foram trabalhados 3 barramentos (com faixas de demanda diferentes), sendo que para cada um utilizou-se da carga de potência ativa e reativa. Os resultados alcançados são dados em função do erro médio absoluto e do erro percentual médio absoluto; além dessas medidas, foi realizada uma análise sobre o fator de potência para os valores reais e previstos / Abstract: In a system of alternating current electricity, generation, transmission and consumption of electricity are divided into active and reactive power. The planning, operation and analysis of these systems are based on estimates of future energy consumption, and in this context are important predictive models of active and reactive load. This dissertation tested forecasting models for short-term active and reactive load models using artificial neural networks. In particular, are implemented and tested many methods of selection enters. The selection of an appropriate subset of variables for inclusion in a system is a vital step in the development of any model. This is particularly important in forecasting models such as artificial neural networks, due to the performance of the final model is strongly dependent on the input variables used. This dissertation developed a model that supports the integration of some techniques for selection (mutual information and partial mutual information) with the aim to facilitate the use of these, as well as, its comparison when applied to certain prediction problems. For the experiments have been worked 3 buses (with different ranges of demand), and for each one used the load active and reactive power. The results obtained are given in function of the mean absolute error and mean absolute percentage error; in addition to these measures, an analysis was made of the power factor for the actual and target values / Mestrado / Engenharia de Computação / Mestre em Engenharia Elétrica

Orientador: Philippe Remy Bernard Devloo / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenharia Civil, Arquitetura e Urbanismo / Made available in DSpace on 2018-08-19T19:01:38Z (GMT). No. of bitstreams: 1 Santos_ThiagoDiasdos_M.pdf: 3063221 bytes, checksum: 5d64512942a835687c68946975bc53ef (MD5) Previous issue date: 2012 / Resumo: No presente trabalho, desenvolveu-se uma biblioteca para geração de redes neurais artificiais (NeuralLib) e aplicou-se a mesma para aproximação do acoplamento de escoamento em poços horizontais com reservatório. A biblioteca NeuralLib foi desenvolvida em linguagem C++. A arquitetura de rede gerada e utilizada foi a Multilayer Perceptron (MLP) com uma única camada oculta. Optouse em gerar 3 arquiteturas com diferentes números de neurônios ocultos com objetivo de analisar o comportamento das MLPs. O algoritmo de treinamento adotado foi o de retropropagação ou backpropagation. A rede neural foi utilizada para mapear o fluxo do reservatório tridimensional para o poço horizontal. O escoamento no poço é simulado utilizando leis constitutivas turbulentas e laminares. Foi elaborada uma técnica para gerar os conjuntos de padrões para o processo de treinamento das MLPs, utilizando para tal as curvas de fluxo do reservatório para o poço provenientes de um modelo tridimensional. As MLPs treinadas foram utilizadas na resolução de um modelo unidimensional fornecendo valores de um parâmetro de fluxo do reservatório. Nesse processo, o modelo unidimensional produziu curvas de fluxo no poço semelhantes aos gerados pelo modelo tridimensional. Os resultados são avaliados com relação ao processo de treinamento das MLPs e com relação às curvas de fluxo e vazão total de produção dos poços / Abstract: In this work, an object-oriented library was developed which implements neural networks (Neural- Lib). The library was used to model the coupling of the fluid flow in a three-dimensional reservoir with a one-dimensional well model. The architecture of the neural network is the Multilayer Perceptron (MLP) with a single hidden layer. Three different architectures with varying number of hidden neurons were tested to evaluate the behaviour of the MLP. The backpropagation algorithm was used to train the network. The neural network was applied to estimate the mass flux from a three dimensional reservoir to a horizontal well. The fluid flow in the horizontal well uses laminar and turbulent constitutive models. A technique was developed to generate a set of patterns which were used to train the MLP's. The MLP's output data is a function which represents the mass flux from the reservoir to the one dimensional well. Using the mass flux function, the pressure function in the horizontal well and well flux were very close to the pressure and flux computed using the three dimensional model. The effectiveness of the neural network was evaluated by comparing cases which were not included in the original training set / Mestrado / Estruturas / Mestre em Engenharia Civil

Efterfrågan och tillhandahållandet av platsberoende tjänster blir allt större vilket i sin tur skapar intresse för billiga och skalbara tekniker i alla möjliga olika miljöer. Särskilt intressant blir tekniker som är lätta att installera på nya platser och vars hårdvarukomponenter är enkla och billiga. I denna rapport presenteras en experimentiell systemteknisk metod för positionsberäkning i inomhusmiljöer, specifikt de som på grund av lokala elektromagnetiska fält, rörliga större föremål eller oregelbundna ytor skapar störningar som gör det svårt att utföra förlitlig positionering. Systemet utgörs av ett antal wifi-routrar samt en signalmottagre kopplad till en dator med systemets mjukvarukomponent installerad. Resultatet bedömdes utifrån en förväntad nivå av korrekthet, närmare bestämt att minst hälften av systemets bedömningar inte har fel med mer än två meter, samt en övre gräns på högst tre meters fel i minst 90 procent av fallen. För att möta målsättningen utrustades mjukvaran med komponenter tänkta att minimera effekten av störningar. Ett Kalmanfilter ger en bättre tolkning av inkommande mätdata medan en för området vanlig estimeringsalgoritm, så kallad Location Fingerprinting, förstärks med en experimentell uppsättning artificiella neurala neuronnät. Som rapporten kommer visa möter systemet som helhet utmaningen och presterar initialt bättre än väntat (hälften av bedömningarna har ett fel på 1,5 meter eller lägre) men även att det beshöver testas i så många nya miljöer som möjligt så att det kan gå att dra slutsatser om dess mer generella användbarhet. / The demand for and supply of location based services (LBS) is constantly growing, which in turn leads to an unquenchable thirst for affordable, scalable localisation solutions in all kinds of surroundings. Technical solutions that are easy to set up at a new location and whose hardware components are simple and affordable, are especially of interest.This paper describes an experimental system designed for positioning a client in particularly challenging indoor environments – wether it's due to local electromagnetic fields, large moving objects or slanted surfaces, basically whatever could create difficulties in radiowave based positioning. This system consists of a number of wifi routers and a signal receiver connected to a computer running the central software component. The results were assessed out of an expected level of accuracy, namely that no more than half of the estimates are off by two meters or more, with an upper limit of no more than 90 percent of the estimates being off by three meters or more. In order to achieve this, the software includes algorithms designed to lessen the effect of signal disruption. A Kalman filter gives the system a better interpretation of sensor data, while the (for the field) common estimation method of Location Fingerprinting gets reinforced by an experimental array of artificial neural networks. As this paper will show, the system will within the initial testing fulfill the set criteria to satisfaction, however it will need future trials in a row of varying environments so as to give an indication of its general usefulness.

For scientific or socio-economic reasons it is often necessary or desirable that biological material be identified. Given that there are an estimated 10 million living organisms on Earth, the identification of biological material can be problematic. Consequently the services of taxonomist specialists are often required. However, if such expertise is not readily available it is necessary to attempt an identification using an alternative method. Some of these alternative methods are unsatisfactory or can lead to a wrong identification. One of the most common problems encountered when identifying specimens is that important diagnostic features are often not easily observed, or may even be completely absent. A number of techniques can be used to try to overcome this problem, one of which, the Self Organizing Map (or SOM), is a particularly appealing technique because of its ability to handle missing data. This thesis explores the use of SOMs as a technique for the identification of indigenous trees of the Acacia species in KwaZulu-Natal, South Africa. The ability of the SOM technique to perform exploratory data analysis through data clustering is utilized and assessed, as is its usefulness for visualizing the results of the analysis of numerical, multivariate botanical data sets. The SOM’s ability to investigate, discover and interpret relationships within these data sets is examined, and the technique’s ability to identify tree species successfully is tested. These data sets are also tested using the C5 and CN2 classification techniques. Results from both these techniques are compared with the results obtained by using a SOM commercial package. These results indicate that the application of the SOM to the problem of biological identification could provide the start of the long-awaited breakthrough in computerized identification that biologists have eagerly been seeking. / Dissertation (MSc)--University of Pretoria, 2011. / Computer Science / unrestricted

Les diatomées sont des algues microscopiques qui sont largement utilisées pour évaluer la qualité écologique des cours d'eau.Les méthodes utilisées se basent sur des modèles simplifiés de biologie des communautés, dans lesquels seules les réponses individuelles des espèces à l'environnement sont prises en compte.Le test de l'importance de processus complémentaires a montré un impact fort des dynamiques de colonisation des espèces, mais un impact négligeable des phénomènes de compétition ou de facilitation.Ces processus impliquent une structure des assemblages bien plus complexe que celle habituellement assumée par les méthodologies de bioindication.L'adaptation et la mise en oeuvre de méthodes de réseaux de neurones et de logique floue nous ont permis de redéfinir des éco-régions françaises et de décrire des relations générales entre les traits biologiques des espèces et l'environnement, tout en prenant mieux en compte cette complexité. / Diatoms are microscopic algae which are widely used to monitor the ecological quality of strems and rivers. The regular methodologies are based on simpllified community models. In these models, only the invidual species responses to environment are accounted for.Testing the importance of complementary processes showed a significant effect of colonization dynamics, but only a slight effect of biotic relationships. These processes led us to considerate a more complex assemblage structure than the one usually assumed by the biomonitoring methodologies.Therefore we implemented both neural networks models and fuzzy logic methodologies, in order to refine French ecoregions and to describe relationships between species traits and environment.

No description available.

Leakage current and process variations are two primary hurdles in modern VLSI design. It depends exponentially on process and environmental parameters and hence small variations in these result in a large spread in leakage current of manufactured dies. Traditionally, Exponential Quadratic(EQ) models have been used to model leakage current as a function of process parameters which can model limited non-linearity and hence become inaccurate for large process variations. Artificial Neural Networks (ANN) have shown great promise in modeling circuit parameters for CAD applications. We model leakage with ANN models which perform better than the EQ models for increased process variations. However, the complex nature of the ANN model, with the standard sigmoidal activation functions, does not allow analytical expressions for its mean and variance for the case of Gaussian process variations. We propose the use of a new activation function that allows us to derive an analytical expression for the mean and a semi-analytical expression for the variance of the ANN based leakage model. To the best of our knowledge this is the first result in this direction. All existing SLA frameworks are closely tied to the EQ leakage model and hence fail to work with sophisticated ANN models. We therefore set up an SLA framework that can efficiently work with these ANN models. Results show that the CDF of leakage current of ISCAS'85 circuits can be predicted accurately with the error in mean and standard deviation, compared to Monte Carlo based simulations, being less than 1\% and 2\% respectively across a range of voltage and temperature values. The complexity of our framework is similar to existing SLA frameworks yet more accurate over a larger range of variations. Ignoring the thermal profile of the chip leads to a gross error of nearly 50\% in the prediction of leakage yield. Our neural network model also includes the voltage and temperature as input parameters, thereby enabling voltage and temperature aware statistical leakage analysis (SLA). Similarly leakage CDF can be predicted across a range of supply and body voltages since they are both part of the model. Our framework used analytical techniques to account for local variations and Monte Carlo techniques for global variations and hence it can also be used for Non-Gaussian global variations.

A human brain has billions of processing elements operating simultaneously; the only practical way to model this computationally is with a massively-parallel computer. A computer on such a significant scale requires hundreds of thousands of interconnected processing elements, a complex environment which requires many levels of monitoring, management and control. Management begins from the moment power is applied and continues whilst the application software loads, executes, and the results are downloaded. This is the story of the research and development of a framework of scalable management tools that support SpiNNaker, a novel computing architecture designed to model spiking neural networks of biologically-significant sizes. This management framework provides solutions from the most fundamental set of power-on self-tests, through to complex, real-time monitoring of the health of the hardware and the software during simulation. The framework devised uses standard tools where appropriate, covering hardware up / down events and capacity information, through to bespoke software developed to provide real-time insight to neural network software operation across multiple levels of abstraction. With this layered management approach, users (or automated agents) have access to results dynamically and are able to make informed decisions on required actions in real-time.