Hydropower maintains its position as the most important source of renewable electric energy in the world. The efficiency of large hydropower plants is unsurpassed, and after more than hundred years of development, the technology is mature and highly reliable. While new hydro resources are currently being developed in Asia and South America, most European countries go through a phase of intense refurbishment and upgrading of existing plants. Challenges faced by the hydropower industry include a knowledge transfer to new generations and the adaptation of unit designs to meet new operational requirements. As with all branches of engineering, the use of computerized design tools has revolutionized the art of hydropower plant design and the analysis of its performance. In the present work, modern tools like coupled field-circuit models and semi-analytic permeance models are used to address different aspects of electromagnetic analysis of generators in large hydropower plants. The results include the presentation of a mathematical model that uses concepts from rotating field theory to determine the air-gap flux density waveform in a hydroelectric generator. The model was succesfully used to evaluate armature voltage harmonics and damper bar currents at no-load and load conditions. A second study is concerned with the importance of losses due to rotational fields in core loss calculations. It is found that dynamic and rotational effects typically increase the total core loss estimates with about 28% in large hydroelectric generators. In a third study, linear models for the calculation of salient pole shoe form factors at an arbitrary level of magnetic loading are presented. The effect of the damper winding configuration on the damping capability of salient-pole generators is then evaluated in a separate study. The predicted impact of the coupling between damper cages on adjacent poles on the damping torque production is verified in a set of experiments.

The topic of this thesis is computation of Mass waveforms, and we consider a number of different cases: Congruence subgroups of the modular group and Dirichlet characters (chapter 1); congruence subgroups and general multiplier systems and real weight (chapter 2); and noncongruence subgroups (chapter 3). In each case we first discuss the necessary theoretical background. We then outline the algorithm and display some of the results obtained by it.

在植被覆蓋的山區中，由於空載雷射掃描可穿透植被間縫隙的特性，有較高機會收集到植被下的地面資訊，因此適合作為製作植被覆蓋地區數值高程模型的資料來源，而在過濾過程中，一般僅利用點雲間的三維位置關係進行幾何過濾，而全波形空載雷射掃描可另外提供點位的波形寬、振幅值、散射截面積以及散射截面積數等波形資料，本研究將透過波形資料分析進行點雲過濾。 首先經最低點採樣後，本研究利用貝氏定理自動分析並計算得到地面點的波形資料的特徵區間範圍，採用振幅值、散射截面積以及散射截面積係數得到的特徵區間範圍開始第一階段的波形資料過濾，完成後再以第二階段的一般幾何過濾濾除剩餘之非地面點，最後的成果將與航測以及只採用幾何過濾時的成果比較。 由研究成果中顯示，不同的植被覆蓋間的單一回波波形資料的差異較明顯，最後回波類似。同一植被覆蓋下的單一回波及最後回波反應不同。而在成果的比較中，本實驗的成果與不採用波形資料輔助的成果大致相同本研究的成果在部分植被覆蓋的區域成果稍差，但透過波形過濾，可將幾何過濾所需計算的點雲數減少許多，可以增進整理過濾的效率。本研究的成果與航測相比時，在植被覆蓋區域較航測成果貼近實際的地面起伏，數值高程模型成果較為正確。 / In mountain areas covered with vegetation, discrete airborne laser scanning is an appropriate technique to produce DEMs for its laser signal is able to reach the ground beneath the vegetation. Once the scanned data was derived, point cloud filtering was performed based on the geometry relationship between the points at the processing stage. With the development of the advanced full-waveform laser scanning system, the additional waveform data has been proved useful for improving the performance of point cloud filtering. This research therefore focused on using the waveform data to extract DEM over vegetation covered area. The amplitude, backscatter cross-section and backscatter cross-section coefficient were the waveform parameters used to do the filtering. After initial waveform analysis was accomplished, an automated method to determine threshold range of each parameter representing ground points was proposed. By applying the thresholds, the original point cloud was filtered. Geometric filtering method was then used to eliminate the remained non-ground points. As a result, the DEM over the target vegetated area was derived. With the comparison against photogrammetric DEM and DEM derived from traditional filtering method, it was demonstrated that the quality of the resultant DEM was improved.

En els darrers quinze anys la velocitat de bit dels sistemes de comunicació ionosfèrica s'ha incrementat significativament, passant dels 75 bps als 14400 bps en un ample de banda de 3 KHz. Les millores experimentades en el modelatge i simulació de canal, el desenvolupament dels sistemes ALE i l'aplicació de tècniques digitals (software radio, modulacions de banda ampla, etc.) són, entre altres, alguns dels motius que han permès aquest increment en les prestacions dels equips, fent-los aptes per comunicacions de llarga distància amb requeriments de velocitat de bit moderada.En particular, la propagació ionosfèrica és una tècnica especialment interessant per enllaços des de l'Antàrtida, on la comunicació amb satèl·lits geostacionaris no és sempre possible.El Departament de Comunicacions i Teoria del Senyal de l'Escola d'Enginyeria La Salle (Universitat Ramon Llull) està involucrat, juntament amb l'Observatori de L'Ebre, en un projecte de recerca que té per objectiu l'estudi de l'enllaç ionosfèric que s'estableix entre la Base Antàrtica Espanyola i Roquetes. L'enllaç, que té una longitud de més de 12700 quilòmetres, ha de servir per transmetre les dades procedents d'un sensor geomagnètic situat a la Base, cap a l'Observatori.Aquesta tesi estudia diversos aspectes d'aquest projecte, principalment pel que fa a la caracterització de canal i al disseny de la capa física adaptada per transmissió eficient de dades.En referència a la caracterització de canal es mostren els resultats d'una campanya de 60 dies de sondeig duta a terme durant l'estiu austral 2006/07 en la que s'han realitzat proves de banda estreta i banda ampla per tal d'obtenir, entre altres, la disponibilitat de l'enllaç, relació senyal a soroll i dispersió temporal i freqüencial.En l'àmbit de la transmissió de dades es proposa, justifica i verifica experimentalment una part d'una capa física basada en espectre eixamplat per seqüència directa especialment adaptada a enllaços a llocs remots: (i) Sense atribució de freqüències; (ii) Amb potència de transmissió relativament baixa; (iii) Infraestructura de radiació limitada. La proposta contempla el funcionament amb relacions SNR negatives, dispersió temporal i freqüencial moderada, robustesa a la interferència i baixa densitat espectral de potència. / En los últimos quince años la velocidad de bit de los sistemas de comunicación ionosférica se ha incrementado significativamente, pasando de los 75 bps a los 14400 bps en un ancho de banda de 3 kHz. Las mejoras experimentadas en el modelado y simulación del canal, el desarrollo de los sistemas ALE y la aplicación de técnicas digitales (software radio, modulaciones de banda ancha, etc.) son, entre otros, algunos de los motivos que han permitido este incremento en las prestaciones de los equipos, haciéndolos aptos para comunicaciones de larga distancia con requerimientos de velocidad de bit moderada.En particular, la propagación ionosférica es una técnica especialmente interesante para enlaces desde la Antártida, donde la comunicación con satélites geostacionarios no es siempre posible.El Departamento de Comunicaciones y Teoría de la Señal de la Escuela de Ingeniería La Salle (Universidad Ramon Llull) está involucrado, junto al Observatorio del Ebro, en un proyecto de investigación que tiene por objetivo el estudio del enlace ionosférico que se establece entre la Base Antártica Española y Roquetes. El enlace, que tiene una longitud de más de 12700 kilómetros, debe servir para transmitir los datos procedentes de un sensor geomagnético situado en la Base, hacia el Observatorio.Esta tesis estudia varios aspectos de este proyecto, principalmente por lo que respecta a la caracterización del canal y al diseño de la capa física adaptada para la transmisión eficiente de datos.En lo referente a la caracterización del canal se muestran los resultados de una campaña de 60 días de sondeo llevada a cabo durante el verano austral 2006/07 en la que se han realizado pruebas de banda estrecha y banda ancha para obtener, entre otros, la disponibilidad del enlace, relación señal a ruido y dispersión temporal y frecuencial.En el ámbito de la transmisión de datos se propone, justifica y verifica experimentalmente una parte de una capa física basada en espectro ensanchado por secuencia directa especialmente adaptada a enlaces a lugares remotos: (i) Sin atribución de frecuencias; (ii) Con potencia de transmisión relativamente baja; (iii) Infraestructura de radiación limitada. La propuesta contempla el funcionamiento con relaciones SNR negativas, dispersión temporal y frecuencial moderada, robustez a la interferencia y baja densidad espectral de potencia. / In the last fifteen years the data rate of the ionospheric communication systems has increased significantly from 75 bps to 14400 bps in a 3 KHz bandwidth. The improvements achieved in the field of channel modeling and simulation, ALE systems and the application of digital techniques (e.g. software radio, wideband modulations, etc.) have permitted the design of new equipment with better performance. As a result, the establishment of low power, very long radio links with moderate data rate has become feasible.In special, the ionospheric propagation technique is very appropriate for radio links from the Antarctica, where the communication to geostationary satellites is not always possible.The Communications and Signal Theory Department of La Salle School of Engineering (Universitat Ramon Llull) and the Observatori de l'Ebre are involved in a research project dealing with the study of an ionospheric radio link between the Spanish Antarctic Base and Roquetes. The link is 12700 Km. long and is going to be used to transmit the low data rate information from a geomagnetic sensor installed in the Spanish Antarctic Base.This thesis has been carried out in the framework of this project, and its goals are related to the channel characterization and the partial design of a physical layer adapted to an efficient transmission of information over that link.Regarding the channel characterization, the results of a 60-days sounding survey corresponding to the 2006/07 southern summer are shown. These results include, among others, link availability, signal to noise ratio and delay and Doppler spread.In the field of data transmission a new physical layer is proposed and experimentally verified, on the basis of the channel measurements and adapted to the requirements and the special conditions in the Antarctica. This physical layer is based on direct sequence spread spectrum and his main characteristics are: (i) Intended for an HF communication without a fixed frequency allocation; (ii) Low emitted power; (iii) Simple radiation infrastructure. The proposal takes into account the operation with negative signal to noise ratios, moderate delay and Doppler spreads, robustness against interference and low power spectral density.

Multiple-input multiple-output (MIMO) radars, which utilize multiple transmitters and receivers to send and receive independent waveforms, have been actively investigated as a next generation radar technology inspired by MIMO techniques in communication theory. Complementary metal-oxide-semiconductor (CMOS) technology offers an opportunity for dramatic cost and size reduction for a MIMO array. However, the resulting formidable signal processing burden has not been addressed properly and remains a challenge. On the other hand, from a block-level point of view, an analog-to-digital converter (ADC) is required for mixed-signal processing to convert analog signals to digital signals, but an ADC occupies a significant portion of a system's budget. Therefore, improvement of an ADC will greatly enhance various trade-offs. This research presents an alternative and viable approach for a MIMO array from a system architecture point of view, and also develops circuit level improvement techniques for an ADC. This dissertation presents a fully-integrated analog pulse compressor (APC) based on an analog matched filter in a mixed signal domain as a key block for the waveform diversity MIMO radar. The performance gain of the proposed system is mathematically presented, and the proposed system is successfully implemented and demonstrated from the block level to the system level using various waveforms. Various figures of merit are proposed to aid system evaluations. This dissertation also presents a low-power ADC based on an asynchronous sample-and-hold multiplying SAR (ASHMSAR) with an enhanced input range dynamic comparator as a key element of a future system. Overall, with the new ADC, a high level of system performance without severe penalty on power consumption is expected. The research in this dissertation provides low-cost and low-power MIMO solutions for a future system by addressing both system issues and circuit issues comprehensively.

Analog-to-digital converter (ADC) is the main workhorse in a digital waveform recorder. Strictly speaking, an ADC is supposed to perform uniformly, irrespective of the characteristics of the signal to be acquired. However, because of certain hardware related inconsistencies, its performance declines, particularly, when acquiring non-repetitive, fast-rising, high frequency signals. The error and distortion contributed due to its declining performance, for the entire range of signals, can be comprehensively characterized by the static and dynamic nonlinearities. Actual testing of ADCs is the only way of estimating these indices. These characteristics reveal information at the microscopic level, such as bit-level aberrations, code transitions, response and settling trends, etc. These tests attain greater significance, when the digitizer is part of a reference measuring or a calibration system, because, the levels of accuracies to be achieved in such a setup may become comparable to the error introduced by the ADC. Hence, testing ADCs is a priority. International and national standards exist for testing digital waveform recorders and ADCs. For several years, the matter related to reducing static test time of high-resolution ADCs was highlighted through many publications. A critical examination of the literature indicates the major schools-of-thought pursued so far, are, (i) refinements to ramp/triangular signal based static testing, (ii) proposals for use of alternative methods and/or test signals for static test, (iii) innovative ways of achieving a relaxation in signal source requirements and, (iv) efforts to combine static and dynamic test into a single test with an appropriate test signal. As a consequence of the literature review, objectives of the thesis were formulated. They attempt to resolve- (i) Conceive a suitable test signal for simultaneous estimation of static and dynamic nonlinearity through a single test (ii) Explore possibility of employing a low-linearity ramp signal to estimate static nonlinearity (iii) Estimating static nonlinearity by exploiting linearity property of a sine signal • In the first part of the thesis, a method is proposed for the concurrent estimation of static and dynamic nonlinearity characteristics of an ADC, with the application of a single test signal. The novelty arises from the fact that the test signal proposed is new, and so is the concept of extracting the static and dynamic nonlinearity from the ADC output. This was achieved by conceiving a test signal, comprising of a high frequency sinusoid (which addresses the dynamic requirement), modulated by a low frequency ramp (which addresses the static requirement). • Static characteristics of an ADC can be determined directly from the histogram-based quasi-static approach by measuring the ADC output, when excited by an ideal ramp/triangular signal of sufficiently low frequency. This approach requires only a fraction of time compared to the conventional DC test, is straightforward, easy to implement, and, in principle is an accepted method as per the revised IEEE-1057. However, the only drawback is that ramp signal sources are not ideal. Thus, nonlinearity present in the ramp signal gets superimposed on the measured ADC characteristics, which renders them, as such, unusable. The second part of the work describes a proposal to get rid of the ramp signal nonlinearity, before it is applied to the ADC. A simple method is presented which employs a low-linearity ramp signal, but yet causes only a fraction of influence on the measured ADC static characteristics. • The third part of the thesis describes a novel method to estimate the actual static characteristics of an ADC using a low frequency sine signal, say, less than 10 Hz, by employing the histogram-based approach. It is based on the well known fact that variation of sine signal is ‘reasonably linear,’ when the angle is small. In the proposed method, the ADC under test has to be ‘fed’ with this ‘linear’ portion of the sine wave. Due to harmonics and offset in input excitation, this ‘linear’ part of the sine signal is marginally different, compared to an ideal ramp signal of equal amplitude. However, since it is a sinusoid, this difference can be accurately determined and later compensated from the measured ADC output. Thus, the corrected ADC output will correspond to the true ADC static nonlinearity. The proposed approach successfully addresses all the three concerns while estimating static linearity, i.e. it is time-efficient, excites all the ADC code-bins reasonably uniformly and tackles the source linearity issue quite nicely. These proposals are novel, simple, easy to implement, time-efficient and importantly static nonlinearity characteristics determined from them are in good agreement with that estimated by the original DC-based technique. Implementation of each method is discussed along with experimental results, for two 8-bit digital oscilloscopes and a 10-bit real time digitizer. Further details are presented in the thesis.

La caractérisation de la proche surface est un enjeu majeur pour l'industrie pétrolière. Lors des acquisitions terrestres et Ocean Bottom Cable (OBC), les couches superficielles généralement altérées ou peu consolidées, présentent des structures géologiques complexes et ont éventuellement des variations topographiques importantes. Les ondes de surface, énergétiques, se propagent dans ce milieu complexe et dominent les sismogrammes, ce qui masque le signal utile pour le traitement sismique classique et rend difficile l'imagerie à la profondeur du réservoir.Il est donc important de pouvoir atténuer ces ondes, éventuellement d'appliquer des corrections statiques et/ou d'amplitude. Ceci qui nécessite une connaissance précise du modèle de vitesse de la proche surface. L'étude de la dispersion des ondes de surface est couramment utilisée en sismologie globale et à l'échelle géotechnique pour évaluer les propriétés des milieux terrestres. Il existe néanmoins des limitations: la mesure de cette dispersion est souvent difficile et les profils de vitesses obtenus sont 1D. A l'échelle pétrolière, l'hypothèse 1D n'est pas toujours adaptée, ce qui motive l'utilisation d'une méthode alternative d'imagerie plus haute résolution, la méthode d'inversion de la forme d'onde (FWI). Cependant, le modèle de vitesse initial doit être assez précis pour éviter le "cycle-skipping" et permettre la convergence vers la solution optimale.Cette étude explore différentes alternatives de fonctions coûts pour résoudre le "cycle-skipping" et diminuer la dépendance de l'inversion à la qualité du modèle initial. En exprimant les fonctions coûts dans le domaine f-k (fréquence-nombre d'onde) et le domaine f-p (fréquence-lenteur), la FWI est plus robuste. A l'aide d'exemples synthétiques, nous démontrons l'efficacité de ces nouvelles approches qui permettent bien de retrouver les variations latérales de vitesses d'onde S.Dans une seconde partie, nous développons une inversion FWI en "layer stripping", adaptée spécifiquement à la physique des ondes de surface. Comme la profondeur de pénétration de ces ondes dépend de leur longueur d'onde, et donc, de leur contenu fréquentiel, nous proposons d'inverser séquentiellement des plus hautes aux plus basses fréquences de ces ondes pour contraindre successivement les couches superficielles jusqu'aux plus profondes. Un fenêtrage selon la distance source-station est également appliqué. Dans un premier temps seules les courtes distances sont inversées, au fur à mesure les données associées à des plus grandes distances sont rajoutées, plus fortement impactées par le "cycle-skipping". Nous démontrons à l'aide d'exemples synthétiques l'avantage de cette méthode par rapport aux méthodes multi-échelles conventionnelles inversant des basses vers les hautes fréquences.Enfin, l'inversion des ondes de surface pour la caractérisation de la proche surface est confrontée à un cas réel. Nous discutons la construction et la pertinence du modèle initial et les difficultés rencontrées lors de l'inversion. / The characterization of the near surface is an important topic for the oil and gas industry. For land and Ocean Bottom Cable (OBC) acquisitions, weathered or unconsolidated top layers, prominent topography and complex shallow structures may make imaging at target depth very difficult. Energetic and complex surface waves often dominate such recordings, masking the signal and challenging conventional seismic processing. Static corrections and the painstaking removal of surface waves are required to obtain viable exploration information.Yet surface waves, which sample the near surface region, are considered as signal on both the engineering and geotechnical scale as well as the global seismology scale. Their dispersive property is conventionally used in surface wave analysis techniques to obtain local shear velocity depth profiles. But limitations such as the picking of dispersion curves and poor lateral resolution have lead to the proposal of Full Waveform Inversion (FWI) as an alternative high resolution technique. FWI can theoretically be used to explain the complete waveforms recoded in seismograms, but FWI with surface waves has its own set of challenges. A sufficiently accurate initial velocity model is required or otherwise cycle-skipping problems will prevent the inversion to converge.This study investigates alternative misfit functions that can overcome cycle-skipping and decrease the dependence on the initial model required. Computing the data-fitting in different domains such as the frequency-wavenumber (f-k) and frequency-slowness (f-p) domains is proposed for robust FWI, and successful results are achieved with a synthetic dataset, in retrieving lateral shear velocity variations.In the second part of this study a FWI layer stripping strategy, specifically adapted to the physics of surface waves is proposed. The penetration of surface waves is dependent on their wavelength, and therefore on their frequency. High-to-low frequency data is therefore sequentially inverted to update top-to-bottom layer depths of the shear velocity model. In addition, near-to-far offsets are considered to avoid cycle-skipping issues. Results with a synthetic dataset show that this strategy is more successful than conventional multiscale FWI in using surface waves to update the shear velocity model.Finally inversion of surface waves for near surface characterization is attempted on a real dataset at the oil and gas exploration scale. The construction of initial models and the difficulties encountered during FWI with real data are discussed.

La déformation des signaux au long de leur trajet de propagation est un des plus importants facteurs qui doivent être considérés à la réception. Ces effets sont dus à des phénomènes comme l’atténuation, la réflexion, la dispersion et le bruit. Alors que les premiers deux phénomènes sont assez facile à surveiller, parce qu’elles affectent l’amplitude, respectivement le retard des signaux, les deux derniers phénomènes sont plus difficiles à contrôler, parce qu’elles changent les paramètres du signal (amplitude, fréquence et phase) de manière totalement dépendante de l’environnement.Dans cette thèse, l’objectif principal est de contribuer à l’analyse des signaux liés aux différents phénomènes physiques, en visant une meilleure compréhension de ces phénomènes, ainsi que l’estimation de leurs paramètres qui sont intéressants de point de vue applicatif. Plusieurs contextes applicatifs ont été investigués dans deux configurations de : active et passive.Pour la configuration active, le premier contexte applicatif consiste en l’étude du phénomène de cavitation dans le domaine de la surveillance de systèmes hydrauliques. La deuxième application de la configuration active est la détection et le suivi des objets immergés sans synchronisation entre les capteurs d’émission et de réception.Pour la configuration passive, nous nous concentrons sur l’analyse des transitoires de pression dans les conduites d’eau en utilisant une méthode non-intrusive ainsi que sur la surveillance des réseaux d’énergie électrique en présence des phénomènes transitoires comme les arcs électriques.Malgré les différences entre les considérations physiques spécifiques à ces applications, nous proposons un modèle mathématique unique pour les signaux issus des deux types de configurations. Le modèle est basé sur l’analyse des récurrences. Avec ce concept, nous proposons une nouvelle approche pour les ondes basées sur l’espace des phases. Cette technique de construction des formes d’ondes présente l’intérêt de conduire à des méthodes de d’investigation active à haut cadence, très utiles pour la surveillance des phénomènes dynamiques.En plus, nous proposons des approches nouvelles pour l’investigation des caractéristiques des signaux. La première est la mesure TDR* (Time Distributed Recurrences) qui quantifie la matrice des récurrences/ distances et qui est utilisée pour la détection des signaux transitoires. La deuxième approche est l’analyse des phases à plusieurs retards et elle est utilisée pour la discrimination entre des signaux avec des paramètres très proches. Finalement, la quantification des lignes diagonales de la matrice des récurrences est proposée comme alternative pour l’analyse des signaux modulés en fréquence.Les travaux présentent les résultats expérimentaux en utilisant les méthodes théorétiques proposées dans cette thèse. Les résultats sont comparés avec des techniques classiques.Des perspectives de ces travaux, tant dans les domaines théorique et qu’applicatif, sont discutés à la fin du mémoire. / Signals’ deformation along their propagation path is among the most important aspect which has to be taken into account at reception. These effects are caused by phenomena like attenuation, reflection, dispersion and noise. Whereas the first two are rather easy to monitor, because they affect the amplitude, respectively the delay, the latter two are more difficult to control, because they change signals’ parameters (amplitude, frequency and phase) in an environment-dependent manner.In this thesis, the main objective is to contribute to the analysis of signals related to different physical phenomena, aiming to better understand them as well as to estimate their parameters that are interesting from application point of view. Different applicative contexts have been investigated in active and passive sensing configurations. For the active part, we mention the monitoring of cavitation phenomena and its characterization for hydraulic system surveillance. The second application of the active sensing is the underwater object detection and tracking without synchronization between sensors. For the passive configuration, we focus on the pressure transient analysis in water pipes investigation with a non-intrusive method and on the surveillance of electrical power systems in the presence of transient phenomena such as electrical arcs.Despite the differences between the physical considerations, we propose a unique mathematical model of the signals issued from the active/passive sensing system used to analyze the considered phenomena. This model is based on the Recurrence Plot Analysis (RPA) method. With this concept, we propose the phase-space based waveform design. This waveform design technique presents the interest to conduct to a high speed sensing methods, very useful to monitor dynamic phenomena.Moreover, we propose new tools for the investigation of the signals characteristics. The first one is the TDR* measure (Time Distributed Recurrences) that quantifies the recurrence/ distance matrix and it is used for the detection of transient signals. The second one is the multi-lag phase analysis using multiple lags and it is successfully used to discriminate between signals with close parameters. Finally, the diagonal lines quantification of RPA matrix is proposed as an alternative for the analysis of modulated signals.Our work presents the experimental results using the proposed theoretical methods introduced by this thesis. The results are compared with classical techniques.The perspectives of this thesis are presented at the end of this paper.

Title: Automated determination of earthquake source parameters Author: Jiří Vackář Department: Department of Geophysics Supervisor: prof. RNDr. Jiří Zahradník, DrSc., Department of Geophysics Abstract: The thesis deals with methods for automated inversion of seismic source parameters. We studied the influence of structure model used and show an ex- ample how the existing model can be improved. We have developed a new, fully automated tool for the centroid moment tensor (CMT) inversion in a Bayesian framework. It includes automated data retrieval from ArcLink server or local data storage. Step-like disturbances are detected using modeling of the distur- bance according to instrument parameters and such components are automati- cally excluded from further processing. Frequency ranges for the filtration and time windows for the inversion are determined automatically due to epicentral distance. Full-waveform inversion is performed in a space-time grid around a provided hypocenter. A data covariance matrix calculated from pre-event noise yields an automated weighting of the station recordings according to their noise levels and also serves as an automated frequency filter suppressing noisy frequency ranges. The method is tested on synthetic and observed data. It is applied on a dataset from the Swiss seismic...

Au cours des dernières années, s’est développé un intérêt tout particulier pour l’optimisation sans dérivée. Ce domaine de recherche se divise en deux catégories: une déterministe et l’autre stochastique. Bien qu’il s’agisse du même domaine, peu de liens ont déjà été établis entre ces deux branches. Cette thèse a pour objectif de combler cette lacune, en montrant comment les techniques issues de l’optimisation déterministe peuvent améliorer la performance des stratégies évolutionnaires, qui font partie des meilleures méthodes en optimisation stochastique. Sous certaines hypothèses, les modifications réalisées assurent une forme de convergence globale, c’est-à-dire une convergence vers un point stationnaire de premier ordre indépendamment du point de départ choisi. On propose ensuite d’adapter notre algorithme afin qu’il puisse traiter des problèmes avec des contraintes générales. On montrera également comment améliorer les performances numériques des stratégies évolutionnaires en incorporant un pas de recherche au début de chaque itération, dans laquelle on construira alors un modèle quadratique utilisant les points où la fonction coût a déjà été évaluée. Grâce aux récents progrès techniques dans le domaine du calcul parallèle, et à la nature parallélisable des stratégies évolutionnaires, on propose d’appliquer notre algorithme pour résoudre un problème inverse d’imagerie sismique. Les résultats obtenus ont permis d’améliorer la résolution de ce problème. / In recent years, there has been significant and growing interest in Derivative-Free Optimization (DFO). This field can be divided into two categories: deterministic and stochastic. Despite addressing the same problem domain, only few interactions between the two DFO categories were established in the existing literature. In this thesis, we attempt to bridge this gap by showing how ideas from deterministic DFO can improve the efficiency and the rigorousness of one of the most successful class of stochastic algorithms, known as Evolution Strategies (ES’s). We propose to equip a class of ES’s with known techniques from deterministic DFO. The modified ES’s achieve rigorously a form of global convergence under reasonable assumptions. By global convergence, we mean convergence to first-order stationary points independently of the starting point. The modified ES’s are extended to handle general constrained optimization problems. Furthermore, we show how to significantly improve the numerical performance of ES’s by incorporating a search step at the beginning of each iteration. In this step, we build a quadratic model using the points where the objective function has been previously evaluated. Motivated by the recent growth of high performance computing resources and the parallel nature of ES’s, an application of our modified ES’s to Earth imaging Geophysics problem is proposed. The obtained results provide a great improvement for the problem resolution.