Automatic age and gender recognition for speech applications is very important for a number of reasons. One of the reasons is that it can improve human-machine interaction. For example, the advertisements can be specialized based on the age and the gender of the person on the phone. It also can help identify suspects in criminal cases or at least it can minimize the number of suspects. Some other uses of this system can be applied for adaptation of waiting queue music where a different type of music can be played according to the person's age and gender. And also using this age and gender recognition system, the statistics about age and gender information for a specific population can be learned. Machine learning is part of artificial intelligence which aims to learn from data. Machine Learning has a long history. But due to some limitations, for ex. , the cost of computation and due to some inefficient algorithms, it was not applied to speech recognition tasks. Only for a decade, researchers started to apply these algorithms to some real world tasks, for ex., speech recognition, computer vision, finance, banking, robotics etc. In this thesis, recognition of age and gender was done using a popular machine learning algorithm and the performance of the system was compared. Also the dataset included real -life examples, so that the system is adaptable to real world applications. To remove the noise and to get the features of speech examples, some digital signal processing techniques were used. Useful speech features that were used in this work were: pitch frequency and cepstral representations. The performance of the age and gender recognition system depends on the speech features used. As the first speech feature, the fundamental frequency was selected. Fundamental frequency is the main differentiating factor between male and female speakers. Also, fundamental frequency for each age group is different. So in order to build age and gender recognition system, fundamental frequency was used. To get the fundamental frequency of speakers, harmonic to sub harmonic ratio method was used. The speech was divided into frames and fundamental frequency for each frame was calculated. In order to get the fundamental frequency of the speaker, the mean value of all the speech frames were taken. It turns out that, fundamental frequency is not only a good discriminator gender, but also it is a good discriminator of age groups simply because there is a distinction between age groups and the fundamental frequencies. Mel Frequency Cepstral Coefficients (MFCC) is a good feature for speech recognition and so it was selected. Using MFCC, the age and gender recognition accuracies were satisfactory. As an alternative to MFCC, Shifted Delta Cepstral (SDC) was used as a speech feature. SDC is extracted using MFCC and the advantage of SDC is that, it is more robust under noisy data. It captures the essential information in noisy speech better. From the experiments, it was seen that SDC did not give better recognition rates because the dataset did not contain too much noise. Lastly, a combination of pitch and MFCC was used to get even better recognition rates. The final fused system has an overall recognition value of 64.20% on ELSDSR [32] speech corpus.

Sågverken i Sverige letar kontinuerligt efter lösningar för att kunna effektivisera produktionen. I denna studie analyseras specifikt Derome Kinnareds sågens timmerplan. Genom en stopptidsanalys, undersökningar av nya tekniska lösningar, muntliga kommunikationer med personal samt arbetsledare och andra företag införskaffades underlag till analys. Åtgärder som kan utföras för att minimera produktionskostnaderna eller öka produktionen på timmerplan är att implementera ny teknik t.ex. fjärrmätning, digitalisera mätbesked med hjälp av SDC:s arkiv, omorganisera personalstyrkan samt införskaffa bemyndigad mätare. Förslagen indikerar att implementering av dem kan förbättra Kinnaredsågens ekonomi. Beroende på hur mycket företaget är villigt att investera kan det effektiviseras i olika grad. I samband med kommunikation med företagsrepresentanter kom man dock fram till att ny teknik kan vara bättre att implementera vid en nybyggnation.

Cette étude est dédiée au développement de piles à combustible de type SOFC en configuration monochambre. L'originalité de ce type de pile SOFC est l'absence de séparation physique entre les compartiments anodique et cathodique. Un mélange d'hydrocarbure et d'air est injecté sur l'ensemble du dispositif comportant l'électrolyte et les deux électrodes de la pile SOFC. Le principe de fonctionnement de ce type de pile repose sur la différence d'activité catalytique entre les deux électrodes. L'anode doit être sélective pour l'oxydation des hydrocarbures et la cathode pour la réduction de l'oxygène. Cette configuration originale permet de s'affranchir des problèmes d'étanchéité des configurations conventionnelles deux atmosphères, ainsi que d'éviter les contraintes sur l'épaisseur de l'électrolyte suivant la disposition des électrodes (coplanaires ou de part et d'autre de l'électrolyte). Cette configuration conduit à des dispositifs simplifiés du point de vue technologique, et donc moins coûteux.<br />Dans cette étude, les électrodes ont été déposées en couches épaisses par la technologie de sérigraphie sur des pastilles support d'électrolyte. Au niveau matériaux, des électrolytes à base de cérine (en particulier GDC et SDC) ont été utilisés afin de diminuer la température de fonctionnement des piles de type SOFC (par rapport à celles avec un électrolyte YSZ). Des électrodes à propriétés catalytiques bien spécifiques vis à vis du mélange gazeux, ont été aussi étudiés : cermets Ni - GDC ou Ni - SDC pour l'anode, cathodes à base de LSM, BSCF ou SSC. Les poudres initiales ainsi que les couches sérigraphiées ont été caractérisées (analyse de phase, microstructure, stabilité chimique, conductivité électrique...) par diverses méthodes physico-chimiques. Différentes piles ont été élaborées et testées dans un réacteur monochambre sous des mélanges air-propane. L'influence des conditions gazeuses (débit, composition) ainsi que de la température de fonctionnement sur les performances des piles a été étudiée. Compte tenu de nos conditions de test, les performances se sont avérées relativement modestes (densité de puissance maximale de l'ordre de 12 mW.cm-2). Néanmoins, cette étude confirme la validité du concept des piles SOFC monochambres, et a aussi permis de développer un savoir-faire au niveau du laboratoire pour poursuivre les études sur cette thématique.

Dialectal Arabic speech research suffers from the lack of labelled resources and standardised orthography. There are three main challenges in dialectal Arabic speech recognition: (i) finding labelled dialectal Arabic speech data, (ii) training robust dialectal speech recognition models from limited labelled data and (iii) evaluating speech recognition for dialects with no orthographic rules. This thesis is concerned with the following three contributions: Arabic Dialect Identification: We are mainly dealing with Arabic speech without prior knowledge of the spoken dialect. Arabic dialects could be sufficiently diverse to the extent that one can argue that they are different languages rather than dialects of the same language. We have two contributions: First, we use crowdsourcing to annotate a multi-dialectal speech corpus collected from Al Jazeera TV channel. We obtained utterance level dialect labels for 57 hours of high-quality consisting of four major varieties of dialectal Arabic (DA), comprised of Egyptian, Levantine, Gulf or Arabic peninsula, North African or Moroccan from almost 1,000 hours. Second, we build an Arabic dialect identification (ADI) system. We explored two main groups of features, namely acoustic features and linguistic features. For the linguistic features, we look at a wide range of features, addressing words, characters and phonemes. With respect to acoustic features, we look at raw features such as mel-frequency cepstral coefficients combined with shifted delta cepstra (MFCC-SDC), bottleneck features and the i-vector as a latent variable. We studied both generative and discriminative classifiers, in addition to deep learning approaches, namely deep neural network (DNN) and convolutional neural network (CNN). In our work, we propose Arabic as a five class dialect challenge comprising of the previously mentioned four dialects as well as modern standard Arabic. Arabic Speech Recognition: We introduce our effort in building Arabic automatic speech recognition (ASR) and we create an open research community to advance it. This section has two main goals: First, creating a framework for Arabic ASR that is publicly available for research. We address our effort in building two multi-genre broadcast (MGB) challenges. MGB-2 focuses on broadcast news using more than 1,200 hours of speech and 130M words of text collected from the broadcast domain. MGB-3, however, focuses on dialectal multi-genre data with limited non-orthographic speech collected from YouTube, with special attention paid to transfer learning. Second, building a robust Arabic ASR system and reporting a competitive word error rate (WER) to use it as a potential benchmark to advance the state of the art in Arabic ASR. Our overall system is a combination of five acoustic models (AM): unidirectional long short term memory (LSTM), bidirectional LSTM (BLSTM), time delay neural network (TDNN), TDNN layers along with LSTM layers (TDNN-LSTM) and finally TDNN layers followed by BLSTM layers (TDNN-BLSTM). The AM is trained using purely sequence trained neural networks lattice-free maximum mutual information (LFMMI). The generated lattices are rescored using a four-gram language model (LM) and a recurrent neural network with maximum entropy (RNNME) LM. Our official WER is 13%, which has the lowest WER reported on this task. Evaluation: The third part of the thesis addresses our effort in evaluating dialectal speech with no orthographic rules. Our methods learn from multiple transcribers and align the speech hypothesis to overcome the non-orthographic aspects. Our multi-reference WER (MR-WER) approach is similar to the BLEU score used in machine translation (MT). We have also automated this process by learning different spelling variants from Twitter data. We mine automatically from a huge collection of tweets in an unsupervised fashion to build more than 11M n-to-m lexical pairs, and we propose a new evaluation metric: dialectal WER (WERd). Finally, we tried to estimate the word error rate (e-WER) with no reference transcription using decoding and language features. We show that our word error rate estimation is robust for many scenarios with and without the decoding features.

Submitted by Cleide Dantas (cleidedantas@ufpa.br) on 2014-05-21T13:00:33Z No. of bitstreams: 2 license_rdf: 23898 bytes, checksum: e363e809996cf46ada20da1accfcd9c7 (MD5) Tese_EmpilhamentoSismicoSuperficie.pdf: 34321354 bytes, checksum: 98bc579e7260095e946f7fa7acc5a244 (MD5) / Rejected by Irvana Coutinho (irvana@ufpa.br), reason: Indexar os assuntos on 2014-08-07T16:11:52Z (GMT) / Submitted by Cleide Dantas (cleidedantas@ufpa.br) on 2014-09-05T13:23:59Z No. of bitstreams: 2 license_rdf: 23898 bytes, checksum: e363e809996cf46ada20da1accfcd9c7 (MD5) Tese_EmpilhamentoSismicoSuperficie.pdf: 34321354 bytes, checksum: 98bc579e7260095e946f7fa7acc5a244 (MD5) / Approved for entry into archive by Irvana Coutinho (irvana@ufpa.br) on 2014-09-18T11:35:24Z (GMT) No. of bitstreams: 2 license_rdf: 23898 bytes, checksum: e363e809996cf46ada20da1accfcd9c7 (MD5) Tese_EmpilhamentoSismicoSuperficie.pdf: 34321354 bytes, checksum: 98bc579e7260095e946f7fa7acc5a244 (MD5) / Made available in DSpace on 2014-09-18T11:35:24Z (GMT). No. of bitstreams: 2 license_rdf: 23898 bytes, checksum: e363e809996cf46ada20da1accfcd9c7 (MD5) Tese_EmpilhamentoSismicoSuperficie.pdf: 34321354 bytes, checksum: 98bc579e7260095e946f7fa7acc5a244 (MD5) Previous issue date: 2001 / CNPq - Conselho Nacional de Desenvolvimento Científico e Tecnológico / ANP - Agência Nacional do Petróleo / O método de empilhamento sísmico por Superfície de Reflexão Comum (ou empilhamento SRC) produz a simulação de seções com afastamento nulo (NA) a partir dos dados de cobertura múltipla. Para meios 2D, o operador de empilhamento SRC depende de três parâmetros que são: o ângulo de emergência do raio central com fonte-receptor nulo (β₀), o raio de curvatura da onda ponto de incidência normal (R_NIP) e o raio de curvatura da onda normal (R_N). O problema crucial para a implementação do método de empilhamento SRC consiste na determinação, a partir dos dados sísmicos, dos três parâmetros ótimos associados a cada ponto de amostragem da seção AN a ser simulada. No presente trabalho foi desenvolvido uma nova sequência de processamento para a simulação de seções AN por meio do método de empilhamento SRC. Neste novo algoritmo, a determinação dos três parâmetros ótimos que definem o operador de empilhamento SRC é realizada em três etapas: na primeira etapa são estimados dois parâmetros (β°₀ e R°_NIP) por meio de uma busca global bidimensional nos dados de cobertura múltipla. Na segunda etapa é usado o valor de β°₀ estimado para determinar-se o terceiro parâmetro (R°_N) através de uma busca global unidimensional na seção AN resultante da primeira etapa. Em ambas etapas as buscas globais são realizadas aplicando o método de otimização Simulated Annealing (SA). Na terceira etapa são determinados os três parâmetros finais (β₀, R_NIP e R_N) através uma busca local tridimensional aplicando o método de otimização Variable Metric (VM) nos dados de cobertura múltipla. Nesta última etapa é usado o trio de parâmetros (β°₀, R°_NIP, R°_N) estimado nas duas etapas anteriores como aproximação inicial. Com o propósito de simular corretamente os eventos com mergulhos conflitantes, este novo algoritmo prevê a determinação de dois trios de parâmetros associados a pontos de amostragem da seção AN onde há intersecção de eventos. Em outras palavras, nos pontos da seção AN onde dois eventos sísmicos se cruzam são determinados dois trios de parâmetros SRC, os quais serão usados conjuntamente na simulação dos eventos com mergulhos conflitantes. Para avaliar a precisão e eficiência do novo algoritmo, este foi aplicado em dados sintéticos de dois modelos: um com interfaces contínuas e outro com uma interface descontinua. As seções AN simuladas têm elevada razão sinal-ruído e mostram uma clara definição dos eventos refletidos e difratados. A comparação das seções AN simuladas com as suas similares obtidas por modelamento direto mostra uma correta simulação de reflexões e difrações. Além disso, a comparação dos valores dos três parâmetros otimizados com os seus correspondentes valores exatos calculados por modelamento direto revela também um alto grau de precisão. Usando a aproximação hiperbólica dos tempos de trânsito, porém sob a condição de R_NIP = R_N, foi desenvolvido um novo algoritmo para a simulação de seções AN contendo predominantemente campos de ondas difratados. De forma similar ao algoritmo de empilhamento SRC, este algoritmo denominado empilhamento por Superfícies de Difração Comum (SDC) também usa os métodos de otimização SA e VM para determinar a dupla de parâmetros ótimos (β₀, R_NIP) que definem o melhor operador de empilhamento SDC. Na primeira etapa utiliza-se o método de otimização SA para determinar os parâmetros iniciais β°₀ e R°_NIP usando o operador de empilhamento com grande abertura. Na segunda etapa, usando os valores estimados de β°₀ e R°_NIP, são melhorados as estimativas do parâmetro R_NIP por meio da aplicação do algoritmo VM na seção AN resultante da primeira etapa. Na terceira etapa são determinados os melhores valores de β°₀ e R°_NIP por meio da aplicação do algoritmo VM nos dados de cobertura múltipla. Vale salientar que a aparente repetição de processos tem como efeito a atenuação progressiva dos eventos refletidos. A aplicação do algoritmo de empilhamento SDC em dados sintéticos contendo campos de ondas refletidos e difratados, produz como resultado principal uma seção AN simulada contendo eventos difratados claramente definidos. Como uma aplicação direta deste resultado na interpretação de dados sísmicos, a migração pós-empilhamento em profundidade da seção AN simulada produz uma seção com a localização correta dos pontos difratores associados às descontinuidades do modelo. / By using an arbitrary source-receiver configuration and without knowledge of the velocity model the recently introduced seismic data stacking method called Common Reflection Surface (CRS) simulates a zero-offset (ZO) section from multi-coverage seismic reflection data. For 2-D acquisition, as by-products provides three normal ray parameters: 1) the emergence angle (β₀); 2) the radius of curvature of the Normal Incidence Point Wave (R_NIP); and 3) the radius of curvature of the Normal Wave (R_N). The CRS stack is based on the hyperbolic traveltime paraxial approximation depending on β₀, R_NIP and R_N. In this thesis is presented a new algorithm of the CRS stack based on two-parameters and one-parameter search strategy combining global and local optimization methods for determine the three parameters that define the stacking surface (or operator). This is performed in three steps: 1) two-parameters search by applying global optimization to determine β₀ and R_NIP; 2) one-parameter search by applying global optimization to determine R_N; and 3) three-parameters search by applying local optimization to determine three parameters, using as initial approximations the parameter triple of the earlier two steps. In the first two steps is used the Simulated Annealing (SA) algorithm and the Variable Metric algorithm is used in the third step. To simulate the conflicting dip events, for each ZO sample where there are interference of intersecting events is determined an additional parameter triple corresponding to a local minimum. The stacking along the respective operator for each particular event allows to simulate their interference in the simulated ZO section by means of their superposition. This new CRS stack algoritm was applied to synthetic data sets providing high-quality simulated ZO sections and high precision determination of the stack parameters in comparison with the forward modeling. Using the hyperbolic traveltime approximation for identical radii of curvature R_NIP = R_N, an algorithm called Common Diffraction Surface (CDS) stack was developed to simulate ZO sections for diffracted waves. In a similar way to the CRS stack procedure, this new algorithm also uses the SA and VM optimization methods to determine the optimal parameter couple (β₀, R_NIP) that define the best CDS operator. The main features of the algorithm are the data normalization, common-offset data, large aperture of the CDS operator and positive search space for R_NIP. The application of the CDS stack algorithm in a synthetic dataset containing reflected and diffracted wavefields provides as main result a simulated ZO section containing diffracted events clearly defined. The post-stack depth migration of the ZO section locates correctly the discontinuities of the second interface.

Solid oxide fuel cell compositionally graded electrolytes could offer the advantage of improving electrical performance and efficiency compared to single-layered or bi-layered yttria stabilized zirconia and samaria doped ceria electrolytes and improving mechanical performance by reducing thermal expansion mismatch stresses compared to bi-layered electrolytes with sharp interfaces. Manufacturing of these graded structures is difficult if implementing conventional wet ceramic techniques. Suspension plasma spraying is an emerging technology that has the potential to rapidly produce thin, dense ceramic layers with no requirement for post deposition heat treatments. However, SPS requires a careful examination of the stability of the feedstock suspensions in order to produce high quality coatings. Optimum suspension formulations with excellent particle dispersion were designed based on rheological and electrostatic stability measurements. These optimized suspensions were used as feedstocks for the fabrication of suspension plasma sprayed compositionally graded YSZ/SDC layers. The feasibility of fabricating graded electrolyte structures was thus demonstrated.

Solid oxide fuel cell compositionally graded electrolytes could offer the advantage of improving electrical performance and efficiency compared to single-layered or bi-layered yttria stabilized zirconia and samaria doped ceria electrolytes and improving mechanical performance by reducing thermal expansion mismatch stresses compared to bi-layered electrolytes with sharp interfaces. Manufacturing of these graded structures is difficult if implementing conventional wet ceramic techniques. Suspension plasma spraying is an emerging technology that has the potential to rapidly produce thin, dense ceramic layers with no requirement for post deposition heat treatments. However, SPS requires a careful examination of the stability of the feedstock suspensions in order to produce high quality coatings. Optimum suspension formulations with excellent particle dispersion were designed based on rheological and electrostatic stability measurements. These optimized suspensions were used as feedstocks for the fabrication of suspension plasma sprayed compositionally graded YSZ/SDC layers. The feasibility of fabricating graded electrolyte structures was thus demonstrated.

No description available.

Die aktuelle Gesundheitsversorgung ist geprägt durch eine steigende Komplexität, die durch die Verzahnung verschiedener medizinischer Bereiche und die Nutzung immer komplexerer, technisch unterstützter Behandlungsmöglichkeiten bedingt ist. Gleichzeitig stehen Kliniken und das medizinische Fachpersonal unter einem hohen Kosten- und Zeitdruck. Für das Erreichen einer optimalen Behandlung des Patienten durch minimalinvasive und mikrochirurgische Eingriffe sind zunehmend Ansätze notwendig, die auf eine Interoperabilität verschiedener Systeme setzen und die Verwendung von zusätzlichen (semi )automatischen Unterstützungssystemen ermöglichen. Diese Arbeit beschäftigt sich mit Herausforderungen bei der Prozessunterstützung im Operationssaal. Dabei liegt der Fokus auf der Aktivitätserfassung und Datenspeicherung als Grundlage zur Erkennung des chirurgischen Prozesses. In einem ersten Schritt wurde ein theoretisches Vorgehensmodell für die intraoperative Prozessunterstützung auf der Basis eines geschlossenen Regelkreises entwickelt. Dabei steht der chirurgische Prozess im Zentrum. In weiteren Schritten wurden ein System zur Erfassung des aktuellen Arbeitsschrittes auf Basis der vorhandenen Videodaten (z.B. Mikroskopie, Ultraschall) sowie eine zentrale Speicherlösung für den Operationssaal entwickelt. Diese zentralen Komponenten sollen es Systemen ermöglichen, während des Eingriffs Daten an die verschiedenen Nutzer bereitzustellen und diese gleichzeitig für eine spätere Dokumentation vorzuhalten. Im weiteren Verlauf der Arbeit werden Ansätze zum Erreichen einer Interoperabilität von Medizingeräten und IT-Systemen im Gesundheitswesen vorgestellt, da technische Systeme neben zusätzlicher Sensorik eine wertvolle Informationsquelle für die Erfassung des aktuellen Prozesses im Operationssaal darstellen. Im Rahmen der Arbeiten des Projektes OR.NET (BMBF, 2012-2016) wurde mit der IEEE 11073-SDC-Standardfamilie eine Möglichkeit zur offenen Vernetzung geschaffen. Konzeptionell integriert werden die Systeme durch eine Beschreibung von Mehrwertdiensten von der einfachen Anzeige von Geräteparametern bis hin zur Teilautomatisierung von technischen Arbeitsschritten auf Basis des Kommunikationsstandards IEEE11073-SDC. Diese wurden basierend auf dieser Basistechnologie gemeinsam mit verschiedenen Projektpartnern entwickelt und mit Klinikern und Klinikbetreibern evaluiert. Die Ergebnisse der Evaluation zeigen, dass durch eine syntaktische und semantische Interoperabilität neue, nutzbringende Funktionen umgesetzt und die Arbeit der verschiedenen Nutzergruppen im Gesundheitssystem effektiv unterstützt werden können.:Inhalt Abstract Abkürzungsverzeichnis Abbildungsverzeichnis Tabellenverzeichnis 1 Einleitung 1.1 Motivation für diese Arbeit 1.2 Zielsetzung der Arbeit 2 Grundlagen der Arbeit 2.1 Aktuelle Operationssäle 2.2 Prozessunterstützung 2.2.1 Workflow und Prozessmodellierung 2.2.2 Workflow in der Medizin 2.2.3 Kontextsensitive Systeme 2.3 Klinische Anwendungsfälle im Rahmen der Arbeit 2.3.1 Intrakranielle Eingriffe am Gehirn 2.3.2 Transsphenoidale Hypophysenadenomentfernung 2.3.3 Sanierende Ohr-OP 3 Vorgehensmodell zur Prozessunterstützung 3.1 Interpretation und Action 3.2 Data analysis und Monitoring 3.3 Mögliche Herangehensweisen bei der Umsetzung einer Prozessunterstützung 4 Erfassung prozessrelevanter Daten im Operationssaal 4.1 Stand der Forschung 4.1.1 Team Assessment und Performanzüberwachung 4.1.2 OP-Dokumentation, Qualitätssicherung und Elektronische Patientenakte (EPA) 4.1.3 Workflow Recognition 4.2 Erkennung der Interaktion zwischen Medizingerät und medizinischem Personal basierend auf der Analyse von Videodaten 4.2.1 Methode 4.2.2 Evaluation 4.2.3 Diskussion 4.3 Erfassung von Informationen durch OP-Integration 4.3.1 Stand der Forschung 4.3.2 Grundlagen des Projektes OR.NET 4.3.3 Zusammenfassung 5 Datenspeicherung im Operationssaal (Surgical Data Recorder und die Erweiterungen auf Basis von OR.NET) 5.1 Surgical Data Recorder 5.1.1 Anforderungsanalyse 5.1.2 Systemkonzept 5.1.3 Evaluationsstudie 5.2 Anpassungen des Datenaufzeichnungskonzeptes im Rahmen des OR.NET-Projekts 5.3 Diskussion und Vergleich der Ansätze 6 Mehrwertdienste auf Basis einer offenen Vernetzung 6.1 Setzen von eingriffsspezifischen Informationen auf den angeschlossenen Geräten 6.2 Anzeige von Informationen im Sichtfeld des Chirurgen 6.3 Mehrwerte durch Datenintegration 6.4 Funktionen mit Nutzung von Prozessinformationen 7 Entwicklung der OP-Demonstratoren und Evaluation der implementierten Mehrwertdienste mit verschiedenen Anwendergruppen 7.1 Anforderungsanalyse für den Leipziger Demonstrator 7.2 Infrastruktur des Demonstrators 7.3 Integrationsszenarien in den Demonstratoren 7.4 Umgesetzte Mehrwertdienste im Leipziger Demonstrator 7.5 Vorgehen bei der Evaluation des Leipziger Demonstrators 7.5.1 Technische Evaluation 7.5.2 Klinische Evaluation 7.6 Ergebnisse der Evaluation 7.6.1 Technische Evaluation 7.6.2 Klinische Evaluation 7.7 Diskussion der Ergebnisse 8 Ausblick auf Weiterentwicklungen der offenen Vernetzung im Operationssaal 9 Zusammenfassung 10 Literatur 11 Anhang 11.1 Modellierung der sanierenden Ohr-OP als EPK inkl. möglicher Vernetzungsszenarien 11.2 Übersicht der umgesetzten Use-Cases im Leipziger Demonstrator 136 11.3 Fragenkatalog OP-Personal 11.4 Fragebogen Betreiber 12 Eigenständigkeitserklärung 13 Eigene Literatur 14 Danksagung

Ce mémoire de thèse présente deux nouvelles approches pour l'analyse et la commande des systèmes non-linéaires complexes, comme les systèmes dynamiques à commutation de la classe des convertisseurs d'énergie électrique. Ces systèmes ont plusieurs modes de fonctionnement et ont un point de fonctionnement désiré qui, en général, n'est le point d'équilibre d'aucun des modes. Dans cette classe de systèmes, la commutation d'un mode de fonctionnement à un autre est commandée selon une loi qui doit être synthétisée. Par conséquent, la synthèse de commande implique l'étude des conditions qui permettent à un cycle limite stable de s'établir au voisinage du point de fonctionnement désiré, puis de la trajectoire de commande qui permet de l'atteindre en respectant les contraintes physiques de comportement (courant maximum supporté par les composants,. . .) ou les contraintes de temps (durée minimum entre deux commutations,. . .). Le cycle limite sera qualifié d'hybride car il est composé de plusieurs dynamiques(deux dans ces travaux).La première méthode développée s'appuie sur les propriétés géométriques des champs de vecteurs et est une extension d'une partie des travaux de thèse de Manon au LAGEP. Une condition nécessaire et suffisante d'existence et de stabilité d'un cycle limite hybride composé d'une séquence de deux modes de fonctionnement dans IR2 est présentée. Ce cycle définit la région finale à atteindre par le système depuis son état initial, par une trajectoire déterminée de manière optimale selon un critère donné (durée totale, énergie dépensée, . . .). La méthode proposée est appliquée aux convertisseurs d'énergie Buck et Buck-Boost alimentant une charge résistive. Une extension à IRn a été proposée et démontrée. Elle est illustrée sur un système non-linéaire dans IR3.La deuxième méthode est développée dans IR2 et basée sur la théorie de Lyapunov, bien connue en automatique pour étudier la stabilité des systèmes non-linéaires et concevoir des commandes stabilisantes.Il s'agit de déterminer par une approche géométrique, une fonction de Lyapunov quadratique commune aux deux modes de fonctionnement du système, qui permette d'obtenir un cycle limite hybride stable le plus proche possible du point de fonctionnement désiré et une commande stabilisante directe des interrupteurs