A aplicação de técnicas de Inteligência artificial em sistemas elétricos de potência permite a resolução de problemas onde soluções tradicionais não mostram bons resultados. Por exemplo, após a ocorrência de uma falta, um sistema automatizado toma todas as medidas necessárias para proteger o sistema e assumir um estado seguro de operação. Nessa ocasião, cabe aos operadores verificar todas as mensagens e dados disponíveis, selecionando aquelas relevantes, para então chegar a uma conclusão sobre o evento ocorrido. Apenas após esse diagnóstico, com base em seu conhecimento, o operador pode ordenar o restabelecimento do sistema ou sua manobra para outro estado. Esse processo é lento e passível de falhas, pois as informações são usualmente conflitantes e confusas, devido à avalanche de eventos, alarmes e alertas decorrentes. Diante desta dificuldade, é interessante dispor de uma ferramenta automática de diagnóstico e apoio na tomada de decisão que torne essa tarefa mais ágil e com menores riscos. Este trabalho aborda a integração e a interoperabilidade de equipamentos de uma subestação elétrica em conformidade com a norma IEC 61850, e a utilização de sistemas inteligentes com lógica Fuzzy executados em um controlador programável moderno. O sistema proposto faz uso de recursos de comunicação da IEC 61850, com mensagens prioritárias padrão GOOSE incluindo valores analógicos de corrente de falta, além de mensagens no padrão TCP/IP MMS. Essas informações, alarmes e eventos, são processadas de forma automática por um motor Fuzzy que permite aproximar a decisão computacional à decisão humana, incluindo zonas de incerteza e lógica ponderada, permitindo respostas mais abstratas do que algo puramente binário. O sistema proposto foi submetido a um grande número de eventos com um simulador em tempo real, com equipamentos reais, tais como IEDs e um controlador programável de alto desempenho. Além disso o sistema usou um esquema de comunicação inovador para o módulo do processador de dados em tempo real (RTDP), inteiramente baseado nos protocolos da IEC 61850. / The application of artificial intelligence techniques in electric power systems allows solving problems where traditional solutions do not show good results. For example, in electric power systems, after the occurence of a fault, an automated system takes all necessary measures to protect the system and assume a safe state of operation. On this occasion, it is for operators to review all messages and data available, selecting those relevant to then come to a conclusion about the event that occurred. Only after this diagnosis, based on his knowledge the operator may order the restoration of the system or their operation to another state. This process is therefore slow and subject to failure, because the information is usually conflicting and confusing because of the avalanche of events, alarms and warnings arising. Faced with this difficulty, it is interesting to have an automatic diagnostic tool and support in decision making that makes this task quicker and with less risk. This paper addresses the integration and interoperability of an electrical substation equipment in accordance with IEC 61850, and the use of intelligent systems Fuzzy logic running on a modern programmable controller. The proposed system makes use of the IEC 61850 communication resources with priority GOOSE messages including standard analog values of fault current, as well as messages on standard TCP / IP MMS. This information, alarms and events, are processed automatically by a fuzzy engine that allows the computational approach to human decision making, including areas of uncertainty and weighted logic, allowing more abstract answers than purely binary. The proposed system has undergone a large number of events with a simulator in real time with actual equipment such as a programmable controller IEDs and high performance. In addition the system used an innovative communication scheme for the module\'s real-time data processor ( RTDP ), based entirely on the IEC 61850 standard protocols such as MMS and GOOSE messages.

Biodiversity, the variety of life on the planet, is declining due to climate change, population and species interactions and as the result f demographic and landscape dynamics. Integrated model-based assessments play a key role in understanding and exploring these complex dynamics and have proven use in conservation planning. Model-based assessments using Species Distribution Models constitute an efficient means of translating limited point data to distribution probability maps for current and future scenarios in support of conservation decision making. The aims of this doctoral study were to investigate; (1) the use of a hybrid genetic programming to build high quality models that handle noisy real-world presence and absence data, (2) the extension of this solution to exploit the parallelism inherent to genetic programming for fast scenario based decision making tasks, and (3) a conceptual framework to share models in the hope of enabling research synthesis. Subsequent to this, the quality of the method, evaluated with the true skill statistic, was examined with two case studies. The first with a dataset obtained by defining a virtual species, and the second with data extracted from the North American Breeding Bird Survey relating to mourning dove (Zenaida macroura). In these studies, the produced models effectively predicted the species distribution up to 30% of error rate both presence and absence samples. The parallel implementation based on a twenty-node c3.xlarge Amazon EC2 StarCluster showed a linear speedup due to the multiple-deme coarse-grained design. The hybrid fuzzy genetic programming algorithm generated under certain consitions during the case studies significantly better transferable models. / Biodiversidade, a variedade de vida no planeta, está em declínio às alterações climáticas, mudanças nas interações das populações e espécies, bem como nas alterações demográficas e na dinâmica de paisagens. Avaliações integradas baseadas em modelo desempenham um papel fundamental na compreensão e na exploração destas dinâmicas complexas e tem o seu uso comprovado no planejamento de conservação da biodiversidade. Os objetivos deste estudo de doutorado foram investigar; (1) o uso de técnicas de programação genética e fuzzy para construir modelos de alta qualidade que lida com presença e ausência de dados ruidosos do mundo real, (2) a extensão desta solução para explorar o paralelismo inerente à programação genética para acelerar tomadas de decisão e (3) um framework conceitual para compartilhar modelos, na expectativa de permitir a síntese de pesquisa. Subsequentemente, a qualidade do método, avaliada com a true skill statistic, foi examinado com dois estudos de caso. O primeiro utilizou um conjunto de dados fictícios obtidos a partir da definição de uma espécie virtual, e o segundo utilizou dados de uma espécie de pomba (Zenaida macroura) obtidos do North American Breeding Bird Survey. Nestes estudos, os modelos foram capazes de predizer a distribuição das espécies maneira correta mesmo utilizando bases de dados com até 30% de erros nas amostras de presença e de ausência. A implementação paralela utilizando um cluster de vinte nós c3.xlarge Amazon EC2 StarCluster, mostrou uma aceleração linear devido ao arquitetura de múltiplos deme de granulação grossa. O algoritmo de programação genética e fuzzy gerada em determinadas condições durante os estudos de caso, foram significativamente melhores na transferência do que os algoritmos do BIOMOD.

As culturas do milho e da soja respondem pela maior parte da produção nacional de grãos, predominando o sistema de plantio direto. Para uma semeadura direta de qualidade, o bom aterramento do sulco é indispensável, pois garante um ambiente adequado às sementes. Neste sentido, é importante estimar a mobilização de solo promovida por uma haste sulcadora estreita durante esta operação. O modelo analítico existente visa representar a mobilização do solo no sistema de plantio convencional. Como consequência, há situações em que este não pode se aplicado, como no caso de hastes sulcadoras estreitas utilizadas em semeadoras de plantio direto. Nestas situações, o mecanismo de falha do solo pode se alterar, assumindo um comportamento não modelado na literatura. Essa pesquisa propõe um modelo fuzzy capaz de representar estas situações, aproveitando conhecimento da teoria de mecânica dos solos e da análise de resultados experimentais. No modelo proposto, parte das regras descrevem situações não abrangidas pelo modelo analítico, as quais foram formuladas a partir da estimativa das prováveis áreas de solo mobilizado. O modelo fuzzy foi testado com dados de experimentos conduzidos durante a pesquisa, em duas condições de granulometria de solo (arenoso e argiloso). O modelo proposto reproduziu as tendências observadas nos dados experimentais, mas superestimou os valores de área observados, sendo esse efeito bem mais intenso para os dados do experimento em solo arenoso. A superestimativa ocorreu devido à soma de diversos fatores. Um deles é a diferença entre as leituras experimentais, as quais consideram apenas o solo realmente movimentado, e a premissa do modelo analítico, que considera toda a área de solo incluindo aquela cisalhada, porém não mobilizada. Outro fator foi devido ao efeito do disco de corte da palha, que pré-cisalha o solo à frente da ferramenta. No ensaio em solo arenoso os valores observados de área de solo mobilizado foram menores que os esperados, intensificando o efeito de superestimativa do modelo fuzzy, sendo que este efeito não representa uma deficiência deste modelo. / Soybean and corn crops account for the majority of Brazilian crop production, predominantly under no-till system. A high quality no-till seeding requires adequate furrow finishing to ensure a suitable environment for seed germination and emergence. Therefore, it is important to estimate furrow soil mobilization promoted by a chisel opener during seeding operation. Existing analytical models aim to represent soil mobilization in conventional tillage. As result, there are situations which the model cannot be applied, such as the case of chisel opener of no-till seeders. Under these situations, soil failure mechanism may change assuming a behavior not modeled by other studies yet. This research proposes a fuzzy model to represent these situations, taking advantage of knowledge obtained from soil mechanics theory and analysis of experimental data. In the proposed model, part of the rules describes situations not covered by the analytical model, which were formulated based on the estimation of possible mobilized soil areas. The fuzzy model was tested using data from experiments for two conditions of soil particle size (sandy and clay). The model simulated trends observed in the experimental data but overestimated observed values of area, which effect was much more intense on sandy soil experiment. The overestimation was due a sum of factors. One is the difference between experimental and modeled data, the first only consider the really mobilized soil, while the last consider the whole soil area, including the sheared but not mobilized area. Another factor was the effect of the straw cutting disc, which pre-shear the soil in front of the tool. In sandy soil experiment, observed values of soil disturbed area were lower than expected, overestimating the effect of fuzzy model, which was not a deficiency of the model.

Submitted by Maria Aparecida (cidazen@gmail.com) on 2017-04-06T14:32:21Z No. of bitstreams: 1 Ariel Soares.pdf: 3082206 bytes, checksum: 46f45860ff1dce76fc8e91ec4f3546f3 (MD5) / Made available in DSpace on 2017-04-06T14:32:21Z (GMT). No. of bitstreams: 1 Ariel Soares.pdf: 3082206 bytes, checksum: 46f45860ff1dce76fc8e91ec4f3546f3 (MD5) Previous issue date: 2017-02-10 / FAPEMA, CNPQ / This research firstly investigates the privacy requirements of users in Mobile Social Networks (MSNs) through a study with 164 Brazilians, which indicated that their requirements are usually dynamic and contextual. Next, the research applies the Situational Computing paradigm to develop a solution to serve them. This solution is called SelPri, developed as proof of concept in the form of a mobile social application to autonomously adapt the privacy settings of posts in MSNs according to the user situation. SelPri uses a conceptual model with fuzzy logic as the basis for constructing an inference engine to identify mobile user situations from the following context information: location, time of the day, day of week, and co-location. SelPri is integrated with Facebook. Additionally, to show the flexibility of the conceptual model, it is also used to construct an inference engine to be used in a different application domain, the mental health. This second inference engine identifies user situations from different context information: it does not use co-location and uses the user activity. The solution originated in the mental health domain is called SituMan. Two experiments were carried out with both solutions, in order to verify the accuracy of the fuzzy inference engine to identify situations, and to evaluate the user satisfaction. The use experience evaluation with SelPri emphasized that the approach to meet the dynamic and contextdependent privacy requirements was well accepted by the participants and proved to be of practical use. The experiments also showed that both solutions were well evaluated with respect to usability. The accuracy evaluations showed a high hit rate of the inference engines to identify situations: ≈94.6% and ≈ 92.04%, for SelPri and SituMan, respectively. / Esta pesquisa primeiramente investiga os requisitos de privacidade de usuários em Redes Sociais Móveis (RSMs) através de um estudo com 164 brasileiros, o qual indicou que seus requisitos são normalmente dinâmicos e contextuais. Em seguida, a pesquisa aplica o paradigma de Computação Situacional para o desenvolvimento de uma solução para atendê-los. Esta solução é chamada de SelPri, desenvolvida como prova de conceito em forma de uma aplicação social móvel para adaptar com autonomia as configurações de privacidade de postagens em RSMs de acordo com a situação do usuário. O SelPri utiliza um modelo conceitual que faz uso de lógica nebulosa como base para a construção de um motor de inferência para identificar as situações de usuários móveis a partir das seguintes informações de contexto: localização, período do dia, dias da semana, e co-localização. O SelPri é implementado integrado ao Facebook. Adicionalmente, para mostrar a flexibilidade do modelo conceitual, ele é também usado para a construção de um motor de inferência para ser utilizado em um domínio de aplicação diferente, o de saúde mental. Esse motor de inferência identifica situações de usuários a partir de informações contextuais diferentes: não utiliza a co-localização e passa a usar a atividade do usuário. A solução originada no domínio de saúde mental é chamada de SituMan. Dois experimentos foram realizados com ambas soluções, em que objetivaram verificar a acurácia do motor de inferência nebulosa para identificação de situações, e avaliar a satisfação do usuário. A avaliação da experiência de uso realizada com o SelPri destacou que a abordagem para atender os requisitos dinâmicos e dependentes de contexto de privacidade teve uma boa aceitação pelos participantes e provou ser de uso prático. As avaliações de experiência de uso também mostraram que ambas soluções foram bem avaliadas com relação a usabilidade. As avaliações de acurácia mostraram uma taxa de acerto elevada dos motores de inferência para identificar situações: ≈94,6% e ≈92,04%, para o SelPri e SituMan, respectivamente

Made available in DSpace on 2017-07-10T17:11:45Z (GMT). No. of bitstreams: 1 DEISY CARDOZO2.pdf: 3855496 bytes, checksum: f8460cbb1dea66b1db54f85149de53c3 (MD5) Previous issue date: 2013-11-29 / Fundação Parque Tecnológico Itaipu / In critical systems such as airplanes, space ships, underwater vehicles, nuclear power plants, among others, failures and faults can bring catastrophic consequences. Therefore, it is required to take into account possibilities of degradation in the performance of its components. This work presents a fault-tolerant control for the propulsion system of the BA-1, an Autonomous Underwater Vehicle (AUV), by using a Fuzzy Allocation Control. This methodology takes advantage of thrusters redundancy, common in autonomous vehicles, to relocate the available thrusters forces in failure situations. That is, it performs an on line reconfiguration of the thrusterforce allocation matrix in the pseudo inverse method. This approach was tested for all the possible cases of horizontal failure of the BA-1 vehicle by means of numerical simulations.Two cases were studied for each failure situation over a linear trajectory in the x axis; the first one studies the case where failures occurs when the vehicle starts its trajectory; and the second one studies the case where the vehicle suffers failures while travelling in its maximum speed.The studies show that the vehicle has an acceptable behavior on most situations, except when it suffers failures of two thrusters which are parallel to the x axis when travelling at maximum speed. However that problem could be solved if the reference trajectory is modified after the failure occurs. In all, the control system developed presents quick responses and an acceptable degradation of performance. / Em sistemas críticos como aeronaves, naves espaciais, veículos subaquáticos, usinas nucleares entre outros, as consequências da falha ou falta de um componente pode ser catastrófica. Portanto, considerar a possibilidade de desempenho degradado por causa de falhas ou faltas em algum dos componentes é um requisito inerente desses sistemas. Este trabalho apresenta um controle tolerante a faltas do sistema de propulsão de um Veículo Autônomo Subaquático, denominado BA-1, utilizando um controle de alocação nebuloso. Esta metodologia aproveita as vantagens da redundância dos propulsores, comum em veículos autônomos, para realocar as forças de propulsão em situações de falta. Assim, este controle realiza uma reconfiguração on line da matriz de alocação de forças dos propulsores utilizada no método da pseudo inversa. Esta proposta foi testada para todos os possíveis casos de falta horizontal do BA-1 mediante simulações numéricas. Foram utilizados dois estudos de casos para uma trajetória reta no o eixo x; o primeiro quando o veículo inicia a sua trajetória com falta nos propulsores, e o se- gundo quando o veículo sofre faltas nos propulsores na sua velocidade máxima de navegação. Os resultados obtidos mostram que o veículo tem um comportamento aceitável em todas as situações excetuando quando o veículo sofre de falta dos propulsores paralelos ao eixo x na sua máxima velocidade, porém poderia ser solucionado modificando a trajetória desejada na ocorrência dessa falta. Contudo o modelo desenvolvido apresenta rapidez nas respostas e desempenho degradado aceitável.

Submitted by Silvana Teresinha Dornelles Studzinski (sstudzinski) on 2016-11-01T15:54:48Z No. of bitstreams: 1 Rogério Pereira de Sousa_.pdf: 946780 bytes, checksum: ceb6c981273e15ecc58fe661bd02a34a (MD5) / Made available in DSpace on 2016-11-01T15:54:48Z (GMT). No. of bitstreams: 1 Rogério Pereira de Sousa_.pdf: 946780 bytes, checksum: ceb6c981273e15ecc58fe661bd02a34a (MD5) Previous issue date: 2016-08-29 / IFTO - Instituto Federal de Educação, Ciência e Tecnologia do Tocantins / A seleção de bovinos leiteiros, através da utilização do sistema de classificação com características lineares de tipo, reflete no ganho de produção, na vida produtiva do animal, na padronização do rebanho, entre outros. Esta pesquisa operacional obteve suas informações através de pesquisas bibliográficas e análise de base de dados de classificações reais. O presente estudo, objetivou a geração de um modelo de classificação de bovinos leiteiros baseado em “true type”, para auxiliar os avaliadores no processamento e análise dos dados, ajudando na tomada de decisão quanto a seleção da vaca para aptidão leiteira, tornando os dados seguros para futuras consultas. Nesta pesquisa, aplica-se métodos computacionais à classificação de vacas leiteiras mediante a utilização mineração de dados e lógica fuzzy. Para tanto, realizou-se a análise em uma base de dado com 144 registros de animais classificados entre as categorias boa e excelente. A análise ocorreu com a utilização da ferramenta WEKA para extração de regras de associação com o algoritmo apriori, utilizando como métricas objetivas, suporte / confiança, e lift para determinar o grau de dependência da regra. Para criação do modelo de decisão com lógica fuzzy, fez-se uso da ferramenta R utilizando o pacote sets. Por meio dos resultados obtidos na mineração de regras, foi possível identificar regras relevantes ao modelo de classificação com confiança acima de 90%, indicando que as características avaliadas (antecedente) implicam em outras características (consequente), com uma confiança alta. Quanto aos resultados obtidos pelo modelo de decisão fuzzy, observa-se que, o modelo de classificação baseado em avaliações subjetivas fica suscetível a erros de classificação, sugerindo então o uso de resultados obtidos por regras de associação como forma de auxílio objetivo na classificação final da vaca para aptidão leiteira. / The selection of dairy cattle through the use of the rating system with linear type traits, reflected in increased production, the productive life of the animal, the standardization of the flock, among others. This operational research obtained their information through library research and basic analysis of actual ratings data. This study aimed to generate a dairy cattle classification model based on "true type" to assist the evaluators in the processing and analysis of data, helping in decision making and the selection of the cow to milk fitness, making the data safe for future reference. In this research, applies computational methods to the classification of dairy cows by using data mining and fuzzy logic. Therefore, we conducted the analysis on a data base with 144 animals records classified between good and excellent categories. Analysis is made with the use of WEKA tool for extraction of association rules with Apriori algorithm, using as objective metrics, support / confidence and lift to determine the degree of dependency rule. To create the decision model with fuzzy logic, it was made use of R using the tool sets package. Through the results obtained in the mining rules, it was possible to identify the relevant rules with confidence classification model above 90%, indicating that the characteristics assessed (antecedent) imply other characteristics (consequent), with a high confidence. As for the results obtained by the fuzzy decision model, it is observed that the classification model based on subjective assessments is susceptible to misclassification, suggesting then the use of results obtained by association rules as a way to aid goal in the final classification cow for dairy fitness

The greening of air transport is the driver for developing technologies to reduce the environmental impact of aviation with the aim of halving the amount of carbon dioxide (COଶ) emitted by air transport, cutting specific emissions of nitrogen oxides (NO୶) by 80% and halving perceived noise by the year 2020. Fuel Cells (FC) play an important role in the new power generation field as inherently clean, efficient and reliable source of power especially when comparing with the traditional fossil-fuel based technologies. The project investigates the feasibility of using an electric hybrid system consisting of a fuel cell and battery to power a small model aircraft (PiperCub J3). In order to meet the desired power requirements at different phases of flight efficiently, a simulation model of the complete system was first developed, consisting of a Proton Exchange Membrane hybrid fuel cell system, 6DoF aircraft model and neural network based controller. The system was then integrated in one simulation environment to run in real-time and finally was also tested in hardware-in-the-loop with real-time control. The control strategy developed is based on a neural network model identification technique; specifically Model Reference Control (MRC), since neural network is well suited to nonlinear systems. To meet the power demands at different phases of flight, the controller controls the battery current and rate of charging/discharging. Three case studies were used to validate and assess the performance of the hybrid system: battery fully charged (high SOC), worst case scenario and taking into account the external factors such as wind speeds and wind direction. In addition, the performance of the Artificial Neural Network Controller was compared to that of a Fuzzy Logic controller. In all cases the fuel cell act as the main power source for the PiperCub J3 aircraft. The tests were carried-out in both simulation and hardware-in-the-loop.

Malgré leur fort développement en assainissement domestique urbain, les bioréacteurs à membranes (BaM) pâtissent de phénomènes de colmatage, induisant des coûts énergétiques et de maintenance importants. L’aération séquencée des membranes par des grosses bulles est l’une des stratégies pour limiter le colmatage ; son impact a fait l’objet de plusieurs études mais reste globalement mal compris, notamment du fait de la complexité de ces systèmes, multiphasiques et opaques. L’hydrodynamique des réacteurs reste mal caractérisée en présence de boues. Pour apporter des éléments de compréhension aux mécanismes de limitation du colmatage par injection d’air, un pilote de filtration membranaire semi-industriel (2 m3, 3 sous modules fibres creuses Puron®) a été conçu, dimensionné et installé sur l’unité de traitement des jus (TDJ) de la station d’épuration Seine Aval (SIAAP), afin d’être alimenté en boues biologiques dans des conditions réelles de fonctionnement. Le suivi des paramètres opératoires du pilote et de ses performances de filtration sur une période de 5 mois avait un double objectif : (i) mieux caractériser la dispersion du gaz pour différentes conditions de fonctionnement (paramètres de l’aération, concentration en boues de l’alimentation), (ii) hiérarchiser les facteurs qui limitent le colmatage des membranes. Il s’agit in fine de proposer des stratégies d’aération adaptées et efficaces pour limiter le colmatage. Afin de caractériser la dispersion du gaz dans le réacteur, une méthodologie innovante basée sur la tomographie de résistivité électrique (ERT) a été adaptée au pilote. Les conditions d’utilisation de l’ERT (nombre d’électrodes de mesure, séquence de quadripôles) ont été sélectionnées à travers une étude numérique, de même que les paramètres d’inversion nécessaires pour reconstituer la cartographie des résistivités à partir des mesures expérimentales. Cette étude numérique poussée, réalisée sous COMSOL, a permis de conclure à l’intérêt de la méthode pour représenter la distribution des phases dans la géométrie considérée. L’ERT a donc été appliquée au pilote alimenté en boues, pour différentes conditions d’aération. Le jeu de données de filtration a par ailleurs été analysé par logique floue, à l’aide du logiciel FisPro. Les arbres de décision obtenus, en analysant les résultats de manière globale et en les regroupant par conditions opératoires similaires, ont mis en évidence l’impact prépondérant des variables suivantes sur la dérive de perméabilité observée (comprise entre - 9 et 2 LMH/bar) : la différence de DCO entre le surnageant des boues et le perméat (DDCO) traduisant une phase colloïdale complexe, et la concentration en matières en suspension (MES), ayant toutes deux un impact négatif sur les performances de filtration. Une augmentation du débit d’air conduirait à une limitation de la dérive de perméabilité, sauf lorsque la variable DCO est élevée (> à 500 mg/L), cette hypothèse restant cependant à vérifier sur une base de données plus conséquente. Le modèle ainsi obtenu par logique floue permet de mieux simuler les évolutions de perméabilité que les modèles obtenus par régression linéaire multivariée (erreurs de 0,61 et de 0,70 respectivement), et ce malgré une incertitude relative importante sur la mesure de perméabilité (jusqu’à 16 %). Ces résultats sont cohérents avec la dispersion du gaz observée par ERT : son homogénéité dépend de la concentration en MES et du débit d’air injecté. A forte concentration en MES (6 – 10 g/L), des zones préférentielles de passage des bulles ont été observées, en particulier à faible débit d’air, expliquant ainsi un colmatage plus important. L’utilisation nouvelle dans ce contexte de ces techniques, ERT et logique floue, donne des résultats qui confortent l’intérêt d’adapter l’aération (débit, séquençage) aux caractéristiques des boues notamment leurs concentrations, et qui permettent d’envisager des stratégies de contrôle de ces paramètres / Membrane bioreactors (MBRs) are widely used in the wastewater treatment sector. However, membrane fouling mitigation remains challenging, and leads to important maintenance and energy costs. Several strategies have been developed industrially to enhance MBR productivity, including coarse bubble sequenced aeration for fouling mitigation. The way such aeration participates in sludge hydrodynamic patterns is an important research topic. However, the methods currently used for hydrodynamic characterisation suffer from several drawbacks, mainly due to the system’s complexity (three phases with opaque deformable solids). More research is needed to characterize hydrodynamics in MBRs filled with activated sludge. A semi-industrial reactor (2.0 m3) was designed and equipped with three hollow fiber membrane modules (KMS Puron) to contribute to fill this knowledge gap. The reactor was continuously fed with activated sludge from one of the reactors of the Seine Aval (SIAAP) wastewater treatment plant. Operating parameters, activated sludge properties as well as process performances were monitored for five months with two objectives: (i) to characterize gas dispersion for different operating conditions (aeration strategy, activated sludge concentrations), (ii) to rank activated sludge properties and operating conditions, according to their impact on fouling mitigation. The ultimate goal is to propose adapted aeration strategies allowing sustainable filtration performance. In order to characterize the gas dispersion in a complex geometry such as MBR membrane tanks, an innovative approach based on electrical resistivity tomography was adapted to the pilot conditions. A numerical approach was used to define the experimental design in terms of electrode positions, quadripole sequences but also inversion parameters, used to reconstruct resistivity maps from experimental datasets. This numerical study, performed on COMSOL, demonstrated the ability of the ERT method to observe different patterns in the membrane zone. ERT was therefore applied to characterise bubble dispersion in the semi-industrial membrane bioreactor filled with activated sludge and operated using different aeration conditions. The dataset obtained from filtration monitoring was analysed using fuzzy logic. The resulting fuzzy decision trees, constructed from the whole dataset or from subsets clustering similar operating conditions, pointed out the impact on the permeability evolution of two main factors: the COD difference between activated sludge supernatant and permeate (COD), standing for complex colloidal phase, and Mixed Liquor Suspended Solid (MLSS) concentration. Both variables emphasizing fouling at high values. Also, an air flow rate augmentation seemed to mitigate fouling, except at high COD concentrations (> 500 mg/L), where its impact was reversed. This last observation needs further investigation, on a larger dataset. The model obtained from fuzzy logic allows a better simulation of permeability evolutions compared to multivariate linear regression (errors of 0.61 and 0.70, respectively), despite a relatively high measurement uncertainty of permeability (up to 16 %). Those results are consistent with the gas dispersion observed with ERT measurements: homogeneity of gas dispersion depends on MLSS concentration and air flow rate. At high concentrations of MLSS (6 g/L and 10 g/L), a high air flow rate is required for the bubbles to be distributed on the whole membrane zone and preferential flow paths have been observed, especially at low air flow rate. This gas dispersion heterogeneity explains worsened filtration performance. In this context, the new use of these technics, ERT and fuzzy logic, provided results that reinforce the interest to link aeration parameters to sludge properties, their MLSS concentration in particular, leading to consider control strategies for these operating parameters.

<p>U ovoj doktorskoj disertaciji predložen je i analiziran metod koji kombinuje primenu entropije odabranih obeležja mrežnog saobraćaja i Takagi-Sugeno-Kang (TSK) neuro-fazi modela u detekciji DoS napada. Entropija je primenjena jer omogućava detekciju širokog spektra statističkih anomalija uzrokovanih DoS napadima dok TSK neuro-fazi model daje dodatni kvalitet u konačnom određivanju tačaka početka i kraja napada povećavajući odnos ispravno i pogrešno detektovanih napada.</p> / <p>In this thesis a new method for DoS attack detection is proposed. This method<br />combines the use of entropy of some characteristic parameters of network traffic<br />and Takagi-Sugeno-Kang (TSK) neuro-fuzzy model. Entropy has been used because<br />it enables detection of wide spectar of network anomalies caused by DoS attacks,<br />while TSK adds new value to final detection of the start and the end of an attack<br />increasing ratio between true and false detections.</p>

This thesis analyses, develops and implements a very fast on-line parameter identification algorithm for both rotor and stator resistances of a rotor flux oriented induction motor drive, with the best possible convergence results using artificial neural networks and fuzzy logic systems. The thesis focuses mainly on identifying the rotor resistance, which is the most critical parameter for RFOC. Limitations of PI and fuzzy logic based estimators were identified. Artificial neural network based estimators were found to track the rotor and stator resistances of the drive accurately and fast. The rotor flux of the induction motor estimated with a classical voltage model was the key input of the rotor resistance estimator. Because, pure digital integrators were unable to play this role, an alternative rotor flux synthesizer using a programmable cascaded filter was developed. This rotor flux synthesizer has been used for all of the resistance estimators. It was found that the error in rotor resistance estimation using an ANN was contributed to by error in the stator resistance (caused by motor heating). Several stator resistance estimators using the stator current measurements were developed. The limitations of a PI and a fuzzy estimator for stator resistance estimation were also established. A new stator resistance identifier using an ANN was found to be much superior to the PI and fuzzy estimators, both in terms of dynamic estimation times and convergence problems. The rotor resistance estimator developed for this thesis used a feedforward neural network and the stator resistance estimator used a recurrent neural network. Both networks exhibited excellent learning capabilities; the stator resistance estimator network was very fast as it had a feedback input. A speed estimator was also developed with the state estimation principles, with the updated motor parameters supplied by the ANN estimators. Analysis for speed sensorless operation has shown that the stator and rotor resistances could be updated on-line.