As pontes de concreto armado são obras de arte de extrema importância para a infraestrutura de transportes do Brasil. Portanto sua inspeção e manutenção são atividades estratégicas. A inspeção visual, ensaios destrutivos e não destrutivos fornecem informações sobre a sua integridade estrutural e auxiliam na tomada de decisões relativas à necessidade de reparos e reforços. Entre os ensaios não destrutivos, avalia-se neste trabalho a aplicação da identificação modal estocástica na detecção de danos em pontes. A técnica baseia-se na medição das vibrações ambientais da estrutura, aquelas que ocorrem durante seu uso, identificação de suas propriedades modais, comparação com as propriedades modais da estrutura íntegra e consequente detecção de danos. Diferentemente da análise dinâmica experimental clássica, na identificação modal estocástica as ações dinâmicas não são medidas e nem controladas durante o ensaio. Por este motivo foram adotadas técnicas de identificação modal baseadas apenas nas vibrações medidas em alguns pontos da estrutura, funções de densidade espectral de potência e transmissibilidades de vibrações entre os pontos. Desta forma as frequências naturais e modos de vibração experimentais puderam ser precisamente identificados em modelos íntegros e danificados de pontes em escala reduzida. Em cada modelo, uma danificação foi imposta em uma de suas longarinas no meio do vão ou no segundo quarto de vão. Após a realização dos ensaios dinâmicos nas condições íntegra e danificada, duas técnicas de identificação de danos foram utilizadas: Diferença de Curvatura Modal (DCM) e Índice de Dano (ID). Ambas as técnicas tiveram sucesso na detecção de danos nos modelos de pontes avaliados. / Reinforced concrete bridges are extremely important elements of Brazilian transportation infrastructure. Consequently their inspection and maintenance are strategic activities. Visual inspection, destructive or nondestructive tests offer relevant information on their structural integrity and support the decision on the need of retrofitting or strengthening. Among existing types of nondestructive tests, this work focuses on the application of stochastic modal identification in damage detection of bridges. This technique is based on the measurement of environmental vibrations that occur during normal operation of the structure, modal identification, comparison of modal properties between damaged and undamaged bridge and finally damage detection. Opposed to classical dynamic experimental analysis, in stochastic modal identification the loads are not measured or known during the test. For this reason modal identification was only based in vibrations measured in selected points of the structure, power spectral density functions and vibration transmissibilities between these points. With this method natural frequencies and experimental modal shapes could be precisely identified in damaged and undamaged small scale models of bridges. The damage was induced in the middle of the span or in the second quarter of the span in one of the girders. After dynamic testing in undamaged and damaged conditions two damage identification techniques were used: Modal Curvature Difference (MCD) and Damage Index (ID). Both techniques detected successfully the damages imposed to the bridge models.

Many of the existing ambient modal identification methods based on vibration data process information centrally to calculate the modal properties. Such methods demand relatively large memory and processing capabilities to interrogate the data. With the recent advances in wireless sensor technology, it is now possible to process information on the sensor itself. The decentralized information so obtained from individual sensors can be combined to estimate the global modal information of the structure. The main objective of this thesis is to present a new class of decentralized algorithms that can address the limitations stated above. The completed work in this regard involves casting the identification problem within the framework of underdetermined blind source separation (BSS). Time-frequency transformations of measurements are carried out, resulting in a sparse representation of the signals. Stationary wavelet packet transform (SWPT) is used as the primary means to obtain a sparse representation in the time-frequency domain. Several partial setups are used to obtain the partial modal information, which are then combined to obtain the global structural mode information. Most BSS methods in the context of modal identification assume that the excitation is white and do not contain narrow band excitation frequencies. However, this assumption is not satisfied in many situations (e.g., pedestrian bridges) when the excitation is a superposition of narrow-band harmonic(s) and broad-band disturbance. Under such conditions, traditional BSS methods yield sources (modes) without any indication as to whether the identified source(s) is a system or an excitation harmonic. In this research, a novel under-determined BSS algorithm is developed involving statistical characterization of the sources which are used to delineate the sources corresponding to external disturbances versus intrinsic modes of the system. Moreover, the issue of computational burden involving an over-complete dictionary of sparse bases is alleviated through a new underdetermined BSS method based on a tensor algebra tool called PARAllel FACtor (PARAFAC) decomposition. At the core of this method, the wavelet packet decomposition coefficients are used to form a covariance tensor, followed by PARAFAC tensor decomposition to separate the modal responses. Finally, the proposed methods are validated using measurements obtained from both wired and wireless sensors on laboratory scale and full scale buildings and bridges.

Many of the existing ambient modal identification methods based on vibration data process information centrally to calculate the modal properties. Such methods demand relatively large memory and processing capabilities to interrogate the data. With the recent advances in wireless sensor technology, it is now possible to process information on the sensor itself. The decentralized information so obtained from individual sensors can be combined to estimate the global modal information of the structure. The main objective of this thesis is to present a new class of decentralized algorithms that can address the limitations stated above. The completed work in this regard involves casting the identification problem within the framework of underdetermined blind source separation (BSS). Time-frequency transformations of measurements are carried out, resulting in a sparse representation of the signals. Stationary wavelet packet transform (SWPT) is used as the primary means to obtain a sparse representation in the time-frequency domain. Several partial setups are used to obtain the partial modal information, which are then combined to obtain the global structural mode information. Most BSS methods in the context of modal identification assume that the excitation is white and do not contain narrow band excitation frequencies. However, this assumption is not satisfied in many situations (e.g., pedestrian bridges) when the excitation is a superposition of narrow-band harmonic(s) and broad-band disturbance. Under such conditions, traditional BSS methods yield sources (modes) without any indication as to whether the identified source(s) is a system or an excitation harmonic. In this research, a novel under-determined BSS algorithm is developed involving statistical characterization of the sources which are used to delineate the sources corresponding to external disturbances versus intrinsic modes of the system. Moreover, the issue of computational burden involving an over-complete dictionary of sparse bases is alleviated through a new underdetermined BSS method based on a tensor algebra tool called PARAllel FACtor (PARAFAC) decomposition. At the core of this method, the wavelet packet decomposition coefficients are used to form a covariance tensor, followed by PARAFAC tensor decomposition to separate the modal responses. Finally, the proposed methods are validated using measurements obtained from both wired and wireless sensors on laboratory scale and full scale buildings and bridges.

As pontes de concreto armado são obras de arte de extrema importância para a infraestrutura de transportes do Brasil. Portanto sua inspeção e manutenção são atividades estratégicas. A inspeção visual, ensaios destrutivos e não destrutivos fornecem informações sobre a sua integridade estrutural e auxiliam na tomada de decisões relativas à necessidade de reparos e reforços. Entre os ensaios não destrutivos, avalia-se neste trabalho a aplicação da identificação modal estocástica na detecção de danos em pontes. A técnica baseia-se na medição das vibrações ambientais da estrutura, aquelas que ocorrem durante seu uso, identificação de suas propriedades modais, comparação com as propriedades modais da estrutura íntegra e consequente detecção de danos. Diferentemente da análise dinâmica experimental clássica, na identificação modal estocástica as ações dinâmicas não são medidas e nem controladas durante o ensaio. Por este motivo foram adotadas técnicas de identificação modal baseadas apenas nas vibrações medidas em alguns pontos da estrutura, funções de densidade espectral de potência e transmissibilidades de vibrações entre os pontos. Desta forma as frequências naturais e modos de vibração experimentais puderam ser precisamente identificados em modelos íntegros e danificados de pontes em escala reduzida. Em cada modelo, uma danificação foi imposta em uma de suas longarinas no meio do vão ou no segundo quarto de vão. Após a realização dos ensaios dinâmicos nas condições íntegra e danificada, duas técnicas de identificação de danos foram utilizadas: Diferença de Curvatura Modal (DCM) e Índice de Dano (ID). Ambas as técnicas tiveram sucesso na detecção de danos nos modelos de pontes avaliados. / Reinforced concrete bridges are extremely important elements of Brazilian transportation infrastructure. Consequently their inspection and maintenance are strategic activities. Visual inspection, destructive or nondestructive tests offer relevant information on their structural integrity and support the decision on the need of retrofitting or strengthening. Among existing types of nondestructive tests, this work focuses on the application of stochastic modal identification in damage detection of bridges. This technique is based on the measurement of environmental vibrations that occur during normal operation of the structure, modal identification, comparison of modal properties between damaged and undamaged bridge and finally damage detection. Opposed to classical dynamic experimental analysis, in stochastic modal identification the loads are not measured or known during the test. For this reason modal identification was only based in vibrations measured in selected points of the structure, power spectral density functions and vibration transmissibilities between these points. With this method natural frequencies and experimental modal shapes could be precisely identified in damaged and undamaged small scale models of bridges. The damage was induced in the middle of the span or in the second quarter of the span in one of the girders. After dynamic testing in undamaged and damaged conditions two damage identification techniques were used: Modal Curvature Difference (MCD) and Damage Index (ID). Both techniques detected successfully the damages imposed to the bridge models.

In the early 1980s, relaying engineers conceived a technology allowing a huge step forward in the monitoring of power system behavior: the synchrophasor, i.e. the estimation of a phasor representation - amplitude and phase - of a sinusoidal waveform at a given point in time thanks to highly accurate time synchronization of a digital relay. By measuring synchrophasors across the power system several times per second, and centralizing the appropriate information in a hierarchical way through a telecommunication network link, it is now possible to continuously monitor the state of very large systems at a high refresh rate. <p><p>At the beginning, the phase angle information of synchrophasors was used to support or improve the performance of classic monitoring applications, such as state estimation and post-mortem analysis. Later, synchrophasors were found to be valuable for the detection and analysis of phenomena that were not monitored previously, such as system islanding and angular stability. This allows a better understanding of system behavior and the design of remedial actions in cases where system security appears to be endangered. Early detection and even prediction of instabilities, as well as validation and improvement of the dynamic models used for studies, have thus become possible.<p><p>However, a power system is rarely stationary and the assumptions behind the definition of “phasor” are not completely fulfilled because the waveforms' frequency and amplitude are not constant over a signal cycle at fundamental frequency. Therefore, accuracy of synchrophasor measurements during dynamic events is an important performance criterion. Furthermore, when discontinuities (phase jumps and high magnitude variations) and harmonics disturb the measured analog signals as a consequence of switching actions or external disturbances, measurements provided to the “user” (the operator or the algorithms that will take decisions such as triggering alarms and remedial actions) require a certain robustness. <p><p>The efforts underpinning this thesis have lead to the development of a method that ensures the robustness of the measurement. This scheme is described and tested in various conditions. In order to achieve a closer alignment between required and actual measurement performance, it is recommended to add an online indicator of phasor accuracy to the phasor data. <p><p>Fast automated corrective actions and closed-loop control schemes relying on synchrophasors are increasingly deployed in power systems. The delay introduced in the measurement and the telecommunication can have a negative impact on the efficiency of these schemes. Therefore, measurement latency is also a major performance indicator of the synchrophasor measurement. <p><p>This thesis illustrates the full measurement chain, from the measurement of analog voltages and currents in the power system to the use of these measurements for various purposes, with an emphasis on real-time applications: visualization, triggering of alarms in the control room or remedial actions, and integration in closed-loop controls. It highlights the various elements along this chain, which influence the availability, accuracy and delay of the data. <p><p>The main focus is on the algorithm to estimate synchrophasors and on the tradeoff between accuracy and latency that arises in applications for which measurements are taken during dynamic events and the data must be processed within a very limited timeframe. <p><p>If both fast phasors and slower, more accurate phasors are made available, the user would be able to select the set of phasors that are the most suitable for each application, by giving priority to either accuracy or a short delay.<p><p>This thesis also tentatively identifies gaps between requirements and typical measurements in order to identify current barriers and challenges to the use of wide area measurement systems. <p><p>A specific application, the continuous monitoring of oscillatory stability, was selected in order to illustrate the benefits of synchrophasors for the monitoring, analysis and control of power system behavior. This application requires a good phasor accuracy but can allow for some measurement delay, unless phasor data are used in an oscillation damping controller. In addition, it also relies on modal estimators, i.e. techniques for the online identification of the characteristics of oscillatory modes from measurements. This field of ongoing research is also introduced in this thesis. / Doctorat en Sciences de l'ingénieur / info:eu-repo/semantics/nonPublished

La robotisation des procédés d’usinage suscite l’intérêt des industriels en raison du grand espace de travail et le faible coût des robots par rapport aux machines-outils conventionnelles et la possibilité d’usiner des pièces de formes complexes. Cependant, la faible rigidité de la structure robotique favorise le déclenchement de phénomènes dynamiques liés à l’usinage sollicitant le robot en bout de l’outil qui dégradent la qualité de surface de la pièce usinée. L’objectif de ces travaux de thèse est de caractériser le comportement dynamique des robots en usinage. Ces travaux ont suivi une démarche en trois étapes : La modélisation d’un premier modèle considéré de référence où le robot est au repos. Ensuite l’identification du comportement dynamique du robot en service. Enfin, l’exploitation des modèles dynamiques du robot en vue de prédire la stabilité de coupe. L’originalité de ces travaux porte sur le développement des méthodes d’identification modale opérationnelles. Elles permettent d’intégrer les conditions réelles d’usinage et d’élaborer des modèles plus précis que le premier modèle de connaissance sans être biaisés par l’effet des harmoniques de rotation de l’outil. Enfin, des préconisations sur le choix de configurations du robot et sur la direction des forces d’excitation sont proposées pour assurer la stabilité de la coupe lors de l’usinage robotisé. / Machining robots have major advantages over cartesian machine tools because of their flexibility, their ability to reach inaccessible areas on a complex part, and their important workspace. However, their lack of rigidity and precision is still a limit for precision tasks. The stresses generated by the cutting forces and inertia are important and cause static and dynamic deformations of the structure which result in problems of workpiece surface. The aim of the thesis work is to characterize the dynamic behavior of robots during machining operation. This work followed a three-step approach : Modeling a first model considered as a reference where the robot is at rest. Then the identification of the dynamic behavior in service. Finally, the prediction of the cutting stability using the robot dynamic model. The originality of this work is the development of new operational modal identification methods. They integrate the machining conditions and result into a more accurate model than the first model of reference without being biased by harmonics. Finally, guidlines of robot’s configurations and excitation forces’ direction are proposed to ensure the robotic machining stability.