De par le fonctionnement de son rotor, l'hélicoptère est le siège de vibrations mécaniques importantes impactant notamment la fatigue des pièces mécaniques et le confort des passagers. La liaison BTP-Fuselage équipé du système SARIB est un système anti-vibratoire qui permet d'atténuer mono-fréquentiellement les vibrations transmises au fuselage. Des solutions intelligentes semi-actives sont donc étudiées afin que la filtration soit réglable en fonction des vibrations excitatrices. Ce type d'études souffre, par contre, d'un manque d'outils et de méthodes indispensables, d'une part, à la modélisation de systèmes mécaniques complexes et d'autre part, à l'élaboration d'une liaison intelligente. Ces travaux proposent une démarche de modélisation à partir d'un outil de modélisation structurel tel que le multi-bond graph (MBG) permettant une vision global et modulaire pour l'étude de systèmes mécaniques complexes tels qu'on peut les trouver sur un hélicoptère. Dans un premier temps, une analyse des outils de modélisation conduisant au choix du MBG a été présentée. Dans un second temps, les développements ont porté sur la modélisation MBG de la liaison BTP/ Fuselage 3D d'un banc d'essai réel qui a été conçu et réalisé au sein du laboratoire. Cette liaison est un système mécanique cinématiquement bouclé. Les équations de la dynamique d'un tel système forment un système d'équations algébro-différentiel (DAE) nécessitant des techniques de résolution spécifiques. Le modèle MBG de la liaison BTP-fuselage entier a été simulé à l'aide du logiciel 20-sim. Les résultats obtenus ont été vérifiés à l'aide du logiciel multicorps LMS Virtual Lab. Une comparaison des résultats obtenus par les deux méthodes a donné, pour différents cas d'excitations de la BTP (pompage, roulis, tangage), une corrélation très satisfaisante. Dans un troisième temps, le modèle MBG a été exploité pour la mise en place d'un dispositif de contrôle semi-actif. Le modèle du dispositif SARIB développé également sous 20-sim permet de régler la position des masses mobiles en fonctionnement de manière à minimiser le niveau de vibratoire du fuselage. L'algorithme de contrôle (algorithme de gradient) permet de calculer les consignes de position des masses mobiles sur les batteurs SARIB. La position des masses mobiles actionnée par un moteur électrique à courant continu et un système vis-écrou est ensuite asservie aux consignes générées par l'algorithme de contrôle. Enfin, la commande a pu être mise en place sur un modèle bond graph non-linéaire qui n'a pas nécessité une linéarisation en vue d'une transformation en fonction de transfert. / Due to the operation of the rotor, the helicopter is subject to important vibrations affecting namely the fatigue of mechanical parts and the passengers comfort. The MGB-Fuselage joint equipped with the DAVI system is an anti-vibration system that helps to reduce, in a single frequency way, vibrations transmitted to the fuselage. Semi-active intelligent solutions are studied so that the filtering can be adjusted according to the vibration sources. Such studies suffer from a lack of tools and necessary methods, firstly, for the design of complex mechanical systems and secondly, for the development of an intelligent joint. This work proposes a modeling approach using a structural modeling tool : the multi-bond graph (MBG) which offers a global and modular view for the study of complex mechatronic systems such as helicopter. At first, an analysis of modeling tools leading to the selection of MBG is presented. Secondly, developments have focused on the MBG modeling of the 3D MGB-fuselage joint of an experimental setup which was designed and built in the laboratory. This joint is a mechanical system with kinematic loops. The equations of the dynamics of such system are a differential-algebraic system (DAE) requiring specific solving methods. The MBG model of the MGB-fuselage was simulated using the 20-sim software. The results were verified using the multibody software LMS Virtual Lab. A comparison of results obtained by the two methods led to a very good correlation to various cases of excitations of the MGB (pumping, roll, pitch). Thirdly, the MBG model was used for the establishment of semi-active control system. The model of the DAVI device also developed in 20-sim allows to adjust the position of the moving masses in operation so as to minimize the level of vibration of the fuselage. The control algorithm (gradient algorithm) enables to calculate the setpoint positions of the moving masses on the DAVI beaters. The position of the moving masses driven by an electric DC motor and a screw-nut system is then controlled to the setpoints generated by the control algorithm. Finally, the command could be implemented on a non-linear bond graph model which did not require a linearization to get a transfer function.

[pt] Desde o final da década de 80 a indústria de fibras óticas têm passado por avanços consideráveis. Através de técnicas controladas, as fibras ópticas podem gerar sinais associados a uma vasta gama de grandezas físicas funcionando como sensores denominados de Sensores a Fibra Óptica (SFO s). Diversas técnicas podem ser empregadas para tal, e entre as existentes a baseada em redes de Bragg é a que mais tem se destacado. O interesse por transdutores empregando esta técnica se justifica pelas vantagens proporcionadas pelo uso da luz, tais como sua capacidade de multiplexação, boa relação sinal/ruído, medições a longas distâncias, imunidade a campos eletromagnéticos, ausência de faísca, entre outras. Neste trabalho buscou-se desenvolver um transdutor e uma técnica de medição baseada em sensores a rede de Bragg para medição de vibrações mecânicas. Um acelerômetro óptico triaxial é projetado e construído. Diferentes modelos foram testados em busca das características de desempenho desejadas. Simulações numéricas empregando o método dos elementos finitos auxiliaram na decisão por melhores desenhos para o transdutor. Resultados de testes experimentais e calibrações empregando um sistema de aquisição de sinais desenvolvido são mostrados. Medições de longa duração para avaliação de estabilidade do sistema e efeitos de temperatura também são apresentados. / [en] Since the end of the 1980s, the fiber optics industry has experienced considerable advances. Through a number of controlled techniques, fiber optics can generate signals associated with a vast array of physical measures, working as sensors denominated Optical Fiber Sensors (OFS s). Many different techniques can be employed to achieve this objective. Among these, the one based on Bragg networks has received the greatest amount of attention. The interest in transducers employing this technique is justified by the advantages of using light, such as its multiplexing capability, good signal-to-noise ratio, possibility of long distance measurements, immunity to electromagnetic fields, and absence of sparks. In the present work, a transducer and measurement technique based on Bragg network sensors vibration are developed, in order to measure mechanical vibrations. A triaxial optical accelerometer is designed and built. Different models are tested in the search for the desired performance characteristics. Numerical simulations employing the finite element method help the decision making process for better transducer designs. Results from experimental and calibration tests using a newly developed signal acquisition system are presented. Long duration measurements to evaluate system stability and temperature effects are also shown.

The thesis is focused on physical principles of signal transmission by optical fiber and effects that may have influence on such transmission. This knowledge is acquired with regard to future usage of optical fiber as a sensor for detection and localization of mechanical vibration. In the thesis, mentioned knowledge is taken in account and also there are described methods, which allow mechanical vibration for dozens of km. At the conclusion the laboratory sollution is suggested, allowing detection and localization in vast distance with possibility of real test in practice.The testing curcuits are used for measurement and results are processed for purpose of detection and localization of source. The measurement itself was made by testing curcuits and results were processed for purpose of detection and localization of source.

O uso de materiais piezelétricos no desenvolvimento de dispositivos para o aproveitamento de energia provinda de vibrações mecânicas, Energy Harvesting, tem sido largamente estudado na última década. Materiais piezelétricos podem ser encontrados na forma de finas camadas ou pastilhas, sendo facilmente integradas a estruturas sem aumento significativo de massa. A conversão de energia mecânica em energia elétrica se dá graças ao acoplamento eletromecânico dos materiais piezelétricos. A maioria das publicações encontradas na literatura exploram o uso de dispositivos eletromecânicos ressonantes, sintonizados na frequência de operação da estrutura, maximizando assim, a energia elétrica de saída dada uma certa condição de operação. O desempenho desses dispositivos ressonantes para coletar e armazenar energia é altamente dependente da adequada sintonização da sua frequência de ressonância com a frequência de operação do sistema/estrutura. Este trabalho apresenta o projeto, otimização e análise de incertezas de um dispositivo coletor/armazenador de energia que consiste em uma placa sob duas condições de contorno, engastada-livre (EL) e deslizante-livre (DL), com massa sísmica e materiais piezelétricos conectados a um circuito shunt. Um modelo em elementos finitos de placa laminada piezelétrica conectada a circuitos R e RL é utilizado combinando as teorias de camada equivalente e deformação de cisalhamento de primeira ordem. A disposição/quantidade de material piezelétrico bem como a massa sísmica acoplados à estrutura foram otimizadas utilizando-se um Algoritmo Genético, levando em conta análises mecânica (modelo mecânico, geometria, peso) e elétrica (modelo elétrico, circuito armazenador). Além disso, o efeito de incertezas dos parâmetros dielétrico e piezelétrico do transdutor, e da indutância elétrica ligada em série ao circuito coletor/armazenador de energia foi estudado. Os resultados indicam que a inclusão de uma indutância sintética ao circuito pode melhorar a coleta de energia em uma banda de frequência e, ainda, que a otimização geométrica pode reduzir a quantidade de material piezelétrico sem no entanto diminuir significativamente a energia gerada. / The use of piezoelectric materials in the development of devices to harvest energy from mechanical vibrations (Energy Harvesting) has been widely studied in the last decade. Piezoelectric materials can be found in the form of thin layers or patches easily integrated into structures without significant mass increase. The conversion of mechanical energy into electric power is provided by the electromechanical coupling of piezoelectric materials. Most publications in the literature explore the use of resonant electromechanical devices, tuned to the operating frequency of the host structure, thus maximizing the power output given a certain operating condition. The performance of these resonant devices to harvest and store energy is highly dependent on the proper tuning of its resonance frequency with the operation frequency of the system/structure. This work presents a design, optimization and uncertainty analysis of energy harvester device consisting of a plate with tip mass and piezoelectric materials connected to shunt circuits. Two boundary conditions are used for the plate, cantilever (EL) and sliding-free (DL). A coupled finite element model with R and RL circuits, combining equivalent single layer and first order shear deformation theories, was used. The distribution and volume of piezoelectric material and the tip mass coupled to the structure were optimized using a Genetic Algorithm, accounting for both mechanical (mechanical model, geometry, weight) and electric (electric model, storer circuit) analyses. Furthermore, the effect of uncertainties of transducer dielectric and piezoelectric constants and electric inductance connected in series with harvesting circuit was studied. The results indicate that the inclusion of a synthetic inductance can improve energy harvesting performance over a frequency range and also that the geometric optimization may reduce the piezoelectric material volume without diminishing significantly the harvested energy.

Vibrações são intrínsecas às máquinas rotativas e, embora não possam ser completamente eliminadas, devem ser controladas de modo a se evitar fadiga e até mesmo falha da máquina. Neste contexto, devido à sua capacidade de alterar as características dinâmicas destas máquinas, os mancais ativos são uma solução efetiva a fim de se reduzir vibrações em rotores, permitindo não só maior ciclo de vida, mas também aumento de confiabilidade e desempenho. Frequentemente, o projeto do sistema de controle destes mancais baseia-se em um modelo matemático da planta, o qual pode ser de difícil obtenção e, devido à adoção de hipóteses simplificadoras (inerentes ao processo de modelagem), pode ser impreciso. Com base nestes conceitos, propõe-se a utilização de uma técnica de controle do tipo proporcional-derivativa baseada em medições de resposta em frequência (livre de modelos matemáticos) aplicada ao controle de vibrações em sistemas rotativos, contornando dificuldades de modelagem. Esta técnica é testada experimentalmente em uma bancada de testes cujos elementos de atuação são os eletromagnetos de um mancal ativo, e um algoritmo para a identificação automática das FRFs do sistema (algoritmo de autoidentificação) é desenvolvido e implementado, permitindo, de forma autônoma, o cálculo dos ganhos ótimos do controlador PD visando atenuação de vibrações. Com base nos resultados obtidos, tem-se que este trabalho é um estudo preliminar que pode viabilizar o desenvolvimento de um mancal ativo inteligente, o qual, a partir de medições do deslocamento do eixo, seria capaz de obter a resposta em frequência do sistema e determinar, de forma automática, os ganhos ótimos do controlador, possibilitando o controle autônomo de vibrações em sistemas rotativos, a partir de um algoritmo de autoidentificação e de uma metodologia de controle livre de modelos. / Vibrations are intrinsic to rotating machinery and, although they cannot be completely eliminated, it is important to control this kind of motion with the objective of avoiding fatigue and even failure of the machine. In this context, due to their capacity of changing the dynamic characteristics of these machines, active bearings are an effective solution to reduce vibration in rotors, allowing not only longer lifecycle, but also higher performance. Frequently, the design of the control system of these bearings is based on a mathematical model of the plant, whose obtainment can be hard and, due to the adoption of simplifying hypotheses (inherent to the modeling process), it may be imprecise. Keeping in mind these concepts, this dissertation proposes the use of a proportional-derivative control technique based on frequency response measurements (free of mathematical models) applied to the vibration control of rotating systems, overcoming modeling difficulties. This technique is experimentally tested in a test rig whose actuation elements are the electromagnets of an active bearing, and an algorithm for automatic identification of the system\'s FRFs (self-identification algorithm) is developed and implemented, allowing, in an autonomous way, the calculation of the optimum gains of the PD controller aiming at controlling vibrations. Based on the obtained results, this work consists in a preliminary study that may enable the development of a smart active bearing, which, from measurements of the shaft\'s displacement, would be capable of obtaining the frequency response of the system and determine, automatically, the optimum gains of the controller, making it possible the autonomous vibration control in rotating systems, from a self-identification algorithm and a model-free control methodology.

Vibrações são intrínsecas às máquinas rotativas e, embora não possam ser completamente eliminadas, devem ser controladas de modo a se evitar fadiga e até mesmo falha da máquina. Neste contexto, devido à sua capacidade de alterar as características dinâmicas destas máquinas, os mancais ativos são uma solução efetiva a fim de se reduzir vibrações em rotores, permitindo não só maior ciclo de vida, mas também aumento de confiabilidade e desempenho. Frequentemente, o projeto do sistema de controle destes mancais baseia-se em um modelo matemático da planta, o qual pode ser de difícil obtenção e, devido à adoção de hipóteses simplificadoras (inerentes ao processo de modelagem), pode ser impreciso. Com base nestes conceitos, propõe-se a utilização de uma técnica de controle do tipo proporcional-derivativa baseada em medições de resposta em frequência (livre de modelos matemáticos) aplicada ao controle de vibrações em sistemas rotativos, contornando dificuldades de modelagem. Esta técnica é testada experimentalmente em uma bancada de testes cujos elementos de atuação são os eletromagnetos de um mancal ativo, e um algoritmo para a identificação automática das FRFs do sistema (algoritmo de autoidentificação) é desenvolvido e implementado, permitindo, de forma autônoma, o cálculo dos ganhos ótimos do controlador PD visando atenuação de vibrações. Com base nos resultados obtidos, tem-se que este trabalho é um estudo preliminar que pode viabilizar o desenvolvimento de um mancal ativo inteligente, o qual, a partir de medições do deslocamento do eixo, seria capaz de obter a resposta em frequência do sistema e determinar, de forma automática, os ganhos ótimos do controlador, possibilitando o controle autônomo de vibrações em sistemas rotativos, a partir de um algoritmo de autoidentificação e de uma metodologia de controle livre de modelos. / Vibrations are intrinsic to rotating machinery and, although they cannot be completely eliminated, it is important to control this kind of motion with the objective of avoiding fatigue and even failure of the machine. In this context, due to their capacity of changing the dynamic characteristics of these machines, active bearings are an effective solution to reduce vibration in rotors, allowing not only longer lifecycle, but also higher performance. Frequently, the design of the control system of these bearings is based on a mathematical model of the plant, whose obtainment can be hard and, due to the adoption of simplifying hypotheses (inherent to the modeling process), it may be imprecise. Keeping in mind these concepts, this dissertation proposes the use of a proportional-derivative control technique based on frequency response measurements (free of mathematical models) applied to the vibration control of rotating systems, overcoming modeling difficulties. This technique is experimentally tested in a test rig whose actuation elements are the electromagnets of an active bearing, and an algorithm for automatic identification of the system\'s FRFs (self-identification algorithm) is developed and implemented, allowing, in an autonomous way, the calculation of the optimum gains of the PD controller aiming at controlling vibrations. Based on the obtained results, this work consists in a preliminary study that may enable the development of a smart active bearing, which, from measurements of the shaft\'s displacement, would be capable of obtaining the frequency response of the system and determine, automatically, the optimum gains of the controller, making it possible the autonomous vibration control in rotating systems, from a self-identification algorithm and a model-free control methodology.

O uso de materiais piezelétricos no desenvolvimento de dispositivos para o aproveitamento de energia provinda de vibrações mecânicas, Energy Harvesting, tem sido largamente estudado na última década. Materiais piezelétricos podem ser encontrados na forma de finas camadas ou pastilhas, sendo facilmente integradas a estruturas sem aumento significativo de massa. A conversão de energia mecânica em energia elétrica se dá graças ao acoplamento eletromecânico dos materiais piezelétricos. A maioria das publicações encontradas na literatura exploram o uso de dispositivos eletromecânicos ressonantes, sintonizados na frequência de operação da estrutura, maximizando assim, a energia elétrica de saída dada uma certa condição de operação. O desempenho desses dispositivos ressonantes para coletar e armazenar energia é altamente dependente da adequada sintonização da sua frequência de ressonância com a frequência de operação do sistema/estrutura. Este trabalho apresenta o projeto, otimização e análise de incertezas de um dispositivo coletor/armazenador de energia que consiste em uma placa sob duas condições de contorno, engastada-livre (EL) e deslizante-livre (DL), com massa sísmica e materiais piezelétricos conectados a um circuito shunt. Um modelo em elementos finitos de placa laminada piezelétrica conectada a circuitos R e RL é utilizado combinando as teorias de camada equivalente e deformação de cisalhamento de primeira ordem. A disposição/quantidade de material piezelétrico bem como a massa sísmica acoplados à estrutura foram otimizadas utilizando-se um Algoritmo Genético, levando em conta análises mecânica (modelo mecânico, geometria, peso) e elétrica (modelo elétrico, circuito armazenador). Além disso, o efeito de incertezas dos parâmetros dielétrico e piezelétrico do transdutor, e da indutância elétrica ligada em série ao circuito coletor/armazenador de energia foi estudado. Os resultados indicam que a inclusão de uma indutância sintética ao circuito pode melhorar a coleta de energia em uma banda de frequência e, ainda, que a otimização geométrica pode reduzir a quantidade de material piezelétrico sem no entanto diminuir significativamente a energia gerada. / The use of piezoelectric materials in the development of devices to harvest energy from mechanical vibrations (Energy Harvesting) has been widely studied in the last decade. Piezoelectric materials can be found in the form of thin layers or patches easily integrated into structures without significant mass increase. The conversion of mechanical energy into electric power is provided by the electromechanical coupling of piezoelectric materials. Most publications in the literature explore the use of resonant electromechanical devices, tuned to the operating frequency of the host structure, thus maximizing the power output given a certain operating condition. The performance of these resonant devices to harvest and store energy is highly dependent on the proper tuning of its resonance frequency with the operation frequency of the system/structure. This work presents a design, optimization and uncertainty analysis of energy harvester device consisting of a plate with tip mass and piezoelectric materials connected to shunt circuits. Two boundary conditions are used for the plate, cantilever (EL) and sliding-free (DL). A coupled finite element model with R and RL circuits, combining equivalent single layer and first order shear deformation theories, was used. The distribution and volume of piezoelectric material and the tip mass coupled to the structure were optimized using a Genetic Algorithm, accounting for both mechanical (mechanical model, geometry, weight) and electric (electric model, storer circuit) analyses. Furthermore, the effect of uncertainties of transducer dielectric and piezoelectric constants and electric inductance connected in series with harvesting circuit was studied. The results indicate that the inclusion of a synthetic inductance can improve energy harvesting performance over a frequency range and also that the geometric optimization may reduce the piezoelectric material volume without diminishing significantly the harvested energy.

Mechanical vibrations in large structures such as buildings, bridges, dams and critical frequencies in large machinery generally have low frequencies (100Hz-1000Hz). To monitor large areas of such structures we need huge network of low cost, easily manufacturable, self-powered and stand-alone vibration spectrum sensors. The sensors should also consume very little power during their overall operation cycle and have moderately high frequency resoultion. The thesis provides mathematical analysis, design and development of stand-alone, low frequency vibration spectrum analyzer .A mechanically stretched polymer piezoelectric membrane, which has a fixed length and tension, can act as a single frequency detector due to its unique resonant frequency. Stretching multiple ribbons of diffferent lengths and tensions, a vibration spectrum analyzer, which gives the Fourier frequency components present in an arbitrary mechanical input vibration, can be designed. The thesis presents a detailed description of experiments to evaluate a low frequency vibration spectrum analyzer system that accepts an incoming input vibration and directly provides the spectrum as output. Polymer piezoelectric materials being easily manufacturable these sensors can be deployed in wide area sensor networks that monitor large structures. The thesis also shows design of a vibration energy harvesting system based on the concept of harvesting energy at low frequencies. The need for developing such an energy harvesting system arises from the necessity of making the vibration sensor, self-powered. Multiple experimental tests were performed before developing a prototype vibration energy harvesting circuit.

In this thesis a new frequency domain approach for the analysis of time delay systems is presented. After linearization of a nonlinear delay differential equation (DDE) with constant distributed delay around a constant or periodic reference solution the so-called Hill-Floquet method can be used for the analysis of the resulting linear DDE. In addition, systems with fast or slowly time-varying delays, systems with variable transport delays originating from a transport with variable velocity, and the corresponding spatially extended systems are presented, which can be also analyzed with the presented method. The newly introduced Hill-Floquet method is based on the Hill’s infinite determinant method and enables the transformation of a system with periodic coefficients to an autonomous system with constant coefficients. This makes the usage of a variety of existing methods for autonomous systems available for the analysis of periodic systems, which implies that the typical calculation of the monodromy matrix for the time evolution of the solution over the principle period is no longer required. In this thesis, the Chebyshev collocation method is used for the analysis of the autonomous systems. Specifically, in this case the periodic part of the solution is expanded in a Fourier series and the exponential behavior of the solution is approximated by the discrete values of the Fourier coefficients at the Chebyshev nodes, whereas in classical spectral or pseudo-spectral methods for the analysis of linear periodic DDEs the complete solution is expanded in terms of basis functions. In the last part of this thesis, new results for three applications with time delay effects are presented, which were analyzed with the presented methods. On the one hand, the occurrence of diffusion-driven instabilities in reaction-diffusion systems with delay is investigated. It is shown that wave instabilities are possible already for single-species reaction diffusion systems with distributed or time-varying delay. On the other hand, the stability of metal cutting vibrations at machine tools is analyzed. In particular, parallel orthogonal turning processes with multiple discrete delays and turning processes with a time-varying delay due to a spindle speed variation are studied. Finally, the stability of the synchronized solution in networks with heterogeneous coupling delays is studied. In particular, the eigenmode expansion for synchronized periodic orbits is derived, which includes an extension of the classical master stability function to networks with heterogeneous coupling delays. Numerical results are shown for a network of Hodgkin-Huxley neurons with two delays in the coupling. / In dieser Dissertation wird ein neues Verfahren zur Analyse von Systemen mit Totzeiten im Frequenzraum vorgestellt. Nach Linearisierung einer nichtlinearen retardierten Differentialgleichung (DDE) mit konstanter verteilter Totzeit um eine konstante oder periodische Referenzlösung kann die sogenannte Hill-Floquet Methode für die Analyse der resultierende linearen DDE angewendet werden. Darüber hinaus werden Systeme mit schnell oder langsam variierender Totzeit, Systeme mit einer variablen Totzeit, resultierend aus einem Transport mit variabler Geschwindigkeit, und entsprechende räumlich ausgedehnte Systeme vorgestellt, welche ebenfalls mit der vorgestellten Methode analysiert werden können. Die neu eingeführte Hill-Floquet Methode basiert auf der Hillschen unendlichen Determinante und ermöglicht die Transformation eines Systems mit periodischen Koeffizienten auf ein autonomes System mit konstanten Koeffizienten. Dadurch können zur Analyse periodischer Systeme auch eine Vielzahl existierender Methoden für autonome Systeme genutzt werden und die Berechnung der Monodromie-Matrix für die Lösung des Systems über eine Periode entfällt. In dieser Arbeit wird zur Analyse des autonomen Systems die Tschebyscheff-Kollokationsmethode verwendet. Im Speziellen wird bei diesem Verfahren der periodische Teil der Lösung in einer Fourierreihe entwickelt und das exponentielle Verhalten durch die Werte der Fourierkoeffizienten an den Tschebyscheff Knoten approximiert, wohingegen bei klassischen spektralen Verfahren die komplette Lösung in bestimmten Basisfunktionen entwickelt wird. Im Anwendungsteil der Arbeit werden neue Ergebnisse für drei Beispielsysteme präsentiert, welche mit den vorgestellten Methoden analysiert wurden. Es wird gezeigt, dass Welleninstabilitäten schon bei Einkomponenten-Reaktionsdiffusionsgleichungen mit verteilter oder variabler Totzeit auftreten können. In einem zweiten Beispiel werden Schwingungen an Werkzeugmaschinen betrachtet, wobei speziell simultane Drehbearbeitungsprozesse und Prozesse mit Drehzahlvariationen genauer untersucht werden. Am Ende wird die Synchronisation in Netzwerken mit heterogenen Totzeiten in den Kopplungstermen untersucht, wobei die Zerlegung in Netzwerk-Eigenmoden für synchrone periodische Orbits hergeleitet wird und konkrete numerische Ergebnisse für ein Netzwerk aus Hodgkin-Huxley Neuronen gezeigt werden.

In this thesis a new frequency domain approach for the analysis of time delay systems is presented. After linearization of a nonlinear delay differential equation (DDE) with constant distributed delay around a constant or periodic reference solution the so-called Hill-Floquet method can be used for the analysis of the resulting linear DDE. In addition, systems with fast or slowly time-varying delays, systems with variable transport delays originating from a transport with variable velocity, and the corresponding spatially extended systems are presented, which can be also analyzed with the presented method. The newly introduced Hill-Floquet method is based on the Hill’s infinite determinant method and enables the transformation of a system with periodic coefficients to an autonomous system with constant coefficients. This makes the usage of a variety of existing methods for autonomous systems available for the analysis of periodic systems, which implies that the typical calculation of the monodromy matrix for the time evolution of the solution over the principle period is no longer required. In this thesis, the Chebyshev collocation method is used for the analysis of the autonomous systems. Specifically, in this case the periodic part of the solution is expanded in a Fourier series and the exponential behavior of the solution is approximated by the discrete values of the Fourier coefficients at the Chebyshev nodes, whereas in classical spectral or pseudo-spectral methods for the analysis of linear periodic DDEs the complete solution is expanded in terms of basis functions. In the last part of this thesis, new results for three applications with time delay effects are presented, which were analyzed with the presented methods. On the one hand, the occurrence of diffusion-driven instabilities in reaction-diffusion systems with delay is investigated. It is shown that wave instabilities are possible already for single-species reaction diffusion systems with distributed or time-varying delay. On the other hand, the stability of metal cutting vibrations at machine tools is analyzed. In particular, parallel orthogonal turning processes with multiple discrete delays and turning processes with a time-varying delay due to a spindle speed variation are studied. Finally, the stability of the synchronized solution in networks with heterogeneous coupling delays is studied. In particular, the eigenmode expansion for synchronized periodic orbits is derived, which includes an extension of the classical master stability function to networks with heterogeneous coupling delays. Numerical results are shown for a network of Hodgkin-Huxley neurons with two delays in the coupling.:1. Introduction 2. System definition and equivalent systems 3. Analysis of nonlinear time delay systems 4. Analytical solution of linear time delay systems 5. Frequency domain approach 6. Hill-Floquet method 7. Applications 8. Concluding remarks A Appendix / In dieser Dissertation wird ein neues Verfahren zur Analyse von Systemen mit Totzeiten im Frequenzraum vorgestellt. Nach Linearisierung einer nichtlinearen retardierten Differentialgleichung (DDE) mit konstanter verteilter Totzeit um eine konstante oder periodische Referenzlösung kann die sogenannte Hill-Floquet Methode für die Analyse der resultierende linearen DDE angewendet werden. Darüber hinaus werden Systeme mit schnell oder langsam variierender Totzeit, Systeme mit einer variablen Totzeit, resultierend aus einem Transport mit variabler Geschwindigkeit, und entsprechende räumlich ausgedehnte Systeme vorgestellt, welche ebenfalls mit der vorgestellten Methode analysiert werden können. Die neu eingeführte Hill-Floquet Methode basiert auf der Hillschen unendlichen Determinante und ermöglicht die Transformation eines Systems mit periodischen Koeffizienten auf ein autonomes System mit konstanten Koeffizienten. Dadurch können zur Analyse periodischer Systeme auch eine Vielzahl existierender Methoden für autonome Systeme genutzt werden und die Berechnung der Monodromie-Matrix für die Lösung des Systems über eine Periode entfällt. In dieser Arbeit wird zur Analyse des autonomen Systems die Tschebyscheff-Kollokationsmethode verwendet. Im Speziellen wird bei diesem Verfahren der periodische Teil der Lösung in einer Fourierreihe entwickelt und das exponentielle Verhalten durch die Werte der Fourierkoeffizienten an den Tschebyscheff Knoten approximiert, wohingegen bei klassischen spektralen Verfahren die komplette Lösung in bestimmten Basisfunktionen entwickelt wird. Im Anwendungsteil der Arbeit werden neue Ergebnisse für drei Beispielsysteme präsentiert, welche mit den vorgestellten Methoden analysiert wurden. Es wird gezeigt, dass Welleninstabilitäten schon bei Einkomponenten-Reaktionsdiffusionsgleichungen mit verteilter oder variabler Totzeit auftreten können. In einem zweiten Beispiel werden Schwingungen an Werkzeugmaschinen betrachtet, wobei speziell simultane Drehbearbeitungsprozesse und Prozesse mit Drehzahlvariationen genauer untersucht werden. Am Ende wird die Synchronisation in Netzwerken mit heterogenen Totzeiten in den Kopplungstermen untersucht, wobei die Zerlegung in Netzwerk-Eigenmoden für synchrone periodische Orbits hergeleitet wird und konkrete numerische Ergebnisse für ein Netzwerk aus Hodgkin-Huxley Neuronen gezeigt werden.:1. Introduction 2. System definition and equivalent systems 3. Analysis of nonlinear time delay systems 4. Analytical solution of linear time delay systems 5. Frequency domain approach 6. Hill-Floquet method 7. Applications 8. Concluding remarks A Appendix