No description available.

Load control strategies for wind turbines are used to reduce the structural wear of the turbine without reducing energy yield. Until now, these control strategies are almost exclusively built up-on linear approaches like PID and model-based controllers due to their robustness. However, advances in turbine size and capabilities create a need for more complex control strategies that can effectively address design challenges in modern turbines. This work presents WINDL, a load control policy based on a neural network, which is trained through model-free Reinforcement Learning (RL) on a simulated wind turbine. While RL has achieved great success in the past on games and simple simulation benchmarks, applications to more complex control problems are starting to emerge just recently. We show that through the usage of regularization techniques and signal transformations, such an application to the field of wind turbine load control is possible. Using a smoothness regularizer, we incentivize the highly non-linear neural network policy to output control actions that are safe to apply to a wind turbine. The Coleman transformation, a common tool for the design of traditional PID-based load control strategies, is used to project signals into a stationary coordinate space, increasing robustness and final policy performance. Trained to control a large offshore turbine in a model-free fashion, WINDL finds a control policy that outperforms a state-of-the-art controller based on the IPC strategy with respect to the prima-ry optimization goal blade loads. Using the DEL metric, we measure 54.1% lower blade loads in the steady wind and 13.45% lower blade loads in the turbulent wind scenario. While such levels of blade reduction come with slightly worse performance on secondary optimi-zation goals like pitch wear and power production, we demonstrate the ability to control the trade-off between different optimization goals on the example of pitch versus blade loads. To comple-ment our findings, we perform a qualitative analysis of the policy behavior and learning process. We believe our work to be the first application of RL to wind turbine load control that exceeds baseline performance in the primary optimization metric, opening up the possibility of including specialized load controllers for targeting critical design driving scenarios of modern large wind turbines.:Problem Method Aim Results Conclusion

[pt] Bandas de cisalhamento ocorrem quando há a localização de deformações inelásticas provenientes de esforços cisalhantes em regiões estreitas de um material. Estas estruturas podem influenciar diretamente nas propriedades dos materiais, além de afetar sua integridade e contribuir para o início de falhas estruturais. Este trabalho apresenta uma metodologia para a caracterização das bandas de cisalhamento na rocha carbonática Indiana Limestone por meio de modelagens numéricas utilizando o método dos elementos finitos (MEF). Ao modelar o fenômeno de localização de deformações, o MEF apresenta algumas limitações como perda da elipticidade das equações governantes, produzindo problemas de convergência e resultados dependentes da discretização de malha. Algumas alternativas para superar estes inconvenientes são apresentadas e discutidas, com especial enfase dada à técnica de regularização viscosa utilizada nas modelagens numericas de ensaios biaxiais e triaxiais. Estudos parametricos e de sensibilidade foram conduzidos para identificar o impacto das propriedades mecânicas na ocorrencia das bandas de cisalhamento. Os resultados mostraram que as propriedades de resistência, o uso de leis de fluxo não associadas e o amolecimento por deformação são os fatores que mais influenciam na iniciação e desenvolvimento das bandas de cisalhamento. / [en] Shear bands occur when inelastic shear deformation localize in narrow regions of the material. These structures can directly influence the properties of materials, in addition to affecting their integrity and contributing to the initiation of structural failures. This study presents a methodology for the characterization of shear bands in Indiana Limestone carbonate rock through numerical modeling using the finite element method (FEM). As it is known, the numerical modeling of strain localization phenomena using FEM has some drawbacks, such as loss of ellipticity of the governing equations, triggering convergence problems and results dependent on the mesh discretization. Some alternatives to overcome these problems are presented and discussed, giving a special emphasis to the viscous regularization technique used in the numerical modeling of biaxial and triaxial tests. Parametric and sensitivity studies were performed to identify the impact of the mechanical properties on the occurrence of shear bands. The results showed that strength properties, non associative flow rules and strain-softening are the factors with larger influence on the initiation and development of shear bands.

[pt] A presente pesquisa tem como foco o desenvolvimento de abordagens híbridas que combinam algoritmos de aprendizado de máquina baseados em conjuntos (ensembles) e técnicas de modelagem e previsão de séries temporais. A pesquisa também inclui o desenvolvimento de heurísticas inteligentes de seleção, isto é, procedimentos capazes de selecionar, dentre o pool de preditores originados por meio dos métodos de conjunto, aqueles com os maiores potenciais de originar previsões agregadas mais acuradas. A agregação de funcionalidades de diferentes métodos visa à obtenção de previsões mais acuradas sobre o comportamento de uma vasta gama de eventos/séries temporais. A tese está dividida em uma sequência de ensaios. Como primeiro esforço, propôsse um método alternativo de geração de conjunto de previsões, o que resultou em previsões satisfatórias para certos tipos de séries temporais de consumo de energia elétrica. A segunda iniciativa consistiu na proposição de uma nova abordagem de previsão combinando algoritmos de Bootstrap Aggregation (Bagging) e técnicas de regularização para se obter previsões acuradas de consumo de gás natural e de abastecimento de energia em diferentes países. Uma nova variante de Bagging, na qual a construção do conjunto de classificadores é feita por meio de uma reamostragem de máxima entropia, também foi proposta. A terceira contribuição trouxe uma série de inovações na maneira pela qual são conduzidas as rotinas de seleção e combinação de modelos de previsão. Os ganhos em acurácia oriundos dos procedimentos propostos são demonstrados por meio de um experimento extensivo utilizando séries das Competições M1, M3 e M4. / [en] This research focuses on the development of hybrid approaches that combine ensemble-based supervised machine learning techniques and time series methods to obtain accurate forecasts for a wide range of variables and processes. It also includes the development of smart selection heuristics, i.e., procedures that can select, among the pool of forecasts originated via ensemble methods, those with the greatest potential of delivering accurate forecasts after aggregation. Such combinatorial approaches allow the forecasting practitioner to deal with different stylized facts that may be present in time series, such as nonlinearities, stochastic components, heteroscedasticity, structural breaks, among others, and deliver satisfactory forecasting results, outperforming benchmarks on many occasions. The thesis is divided into a series of essays. The first endeavor proposed an alternative method to generate ensemble forecasts which delivered satisfactory forecasting results for certain types of electricity consumption time series. In a second effort, a novel forecasting approach combining Bootstrap aggregating (Bagging) algorithms, time series methods and regularization techniques was introduced to obtain accurate forecasts of natural gas consumption and energy supplied series across different countries. A new variant of Bagging, in which the set of classifiers is built by means of a Maximum Entropy Bootstrap routine, was also put forth. The third contribution brought a series of innovations to model selection and model combination in forecasting routines. Gains in accuracy for both point forecasts and prediction intervals were demonstrated by means of an extensive empirical experiment conducted on a wide range of series from the M- Competitions.

Cette thèse concerne le problème de désocclusion d'images, au moyen des équations aux dérivées partielles. Dans la première partie de la thèse, la désocclusion est modélisée par un problème de Cauchy qui consiste à déterminer une solution d'une équation aux dérivées partielles avec des données aux bords accessibles seulement sur une partie du bord de la partie à recouvrir. Ensuite, on a utilisé des algorithmes de minimisation issus de la théorie des jeux, pour résoudre ce problème de Cauchy. La deuxième partie de la thèse est consacrée au choix des paramètres de régularisation pour des EDP d'ordre deux et d'ordre quatre. L'approche développée consiste à construire une famille de problèmes d'optimisation bien posés où les paramètres sont choisis comme étant une fonction variable en espace. Ceci permet de prendre en compte les différents détails, à différents échelles dans l'image. L'apport de la méthode est de résoudre de façon satisfaisante et objective, le choix du paramètre de régularisation en se basant sur des indicateurs d'erreur et donc le caractère à posteriori de la méthode (i.e. indépendant de la solution exacte, en générale inconnue). En outre, elle fait appel à des techniques classiques d'adaptation de maillage, qui rendent peu coûteuses les calculs numériques. En plus, un des aspects attractif de cette méthode, en traitement d'images est la récupération et la détection de contours et de structures fines. / Image inpainting refers to the process of restoring a damaged image with missing information. Different mathematical approaches were suggested to deal with this problem. In particular, partial differential diffusion equations are extensively used. The underlying idea of PDE-based approaches is to fill-in damaged regions with available information from their surroundings. The first purpose of this Thesis is to treat the case where this information is not available in a part of the boundary of the damaged region. We formulate the inpainting problem as a nonlinear boundary inverse problem for incomplete images. Then, we give a Nash-game formulation of this Cauchy problem and we present different numerical which show the efficiency of the proposed approach as an inpainting method.Typically, inpainting is an ill-posed inverse problem for it most of PDEs approaches are obtained from minimization of regularized energies, in the context of Tikhonov regularization. The second part of the thesis is devoted to the choice of regularization parameters in second-and fourth-order energy-based models with the aim of obtaining as far as possible fine features of the initial image, e.g., (corners, edges, … ) in the inpainted region. We introduce a family of regularized functionals with regularization parameters to be selected locally, adaptively and in a posteriori way allowing to change locally the initial model. We also draw connections between the proposed method and the Mumford-Shah functional. An important feature of the proposed method is that the investigated PDEs are easy to discretize and the overall adaptive approach is easy to implement numerically.

Les modèles d’apprentissage profond à large capacité ont souvent tendance à présenter de hauts écarts de généralisation lorsqu’ils sont entrainés avec une quantité limitée de données étiquetées. Dans ce cas, des réseaux de neurones très profonds et larges auront tendance à mémoriser les échantillons de données et donc ils risquent d’être vulnérables lors d’un léger décalage dans la distribution des données au moment de tester. Ce problème produit une généralisation pauvre lors de changements dans la répartition des données au moment du test. Pour surmonter ce problème, certaines méthodes basées sur la dépendance et l’indépendance de données ont été proposées. Une récente classe de méthodes efficaces pour aborder ce problème utilise plusieurs manières de contruire un nouvel échantillon d’entrainement, en mixant une paire (ou plusieurs) échantillons d’entrainement. Dans cette thèse, nous introduisons PatchUp, une régularisation de l’espace des caractéristiques au niveau des blocs dépendant des données qui opère dans l’espace caché en masquant des blocs contigus parmi les caractéristiques mappées, sélectionnés parmi une paire aléatoire d’échantillons, puis en mixant (Soft PatchUp) ou en échangeant (Hard PatchUp) les blocs contigus sélectionnés. Notre méthode de régularisation n’ajoute pas de surcharge de calcul significative au CNN pendant l’entrainement du modèle. Notre approche améliore la robustesse des modèles CNN face au problème d’intrusion du collecteur qui pourrait apparaitre dans d’autres approches de mixage telles que Mixup et CutMix. De plus, vu que nous mixons des blocs contigus de caractéristiques dans l’espace caché, qui a plus de dimensions que l’espace d’entrée, nous obtenons des échantillons plus diversifiés pour entrainer vers différentes dimensions. Nos expériences sur les ensembles de données CIFAR-10, CIFAR-100, SVHN et Tiny-ImageNet avec des architectures ResNet telles que PreActResnet18, PreActResnet34, WideResnet-28-10, ResNet101 et ResNet152 montrent que PatchUp dépasse ou égalise les performances de méthodes de régularisation pour CNN considérée comme état de l’art actuel. Nous montrons aussi que PatchUp peut fournir une meilleure généralisation pour des transformations affines d’échantillons et est plus robuste face à des attaques d’exemples contradictoires. PatchUp aide aussi les modèles CNN à produire une plus grande variété de caractéristiques dans les blocs résiduels en comparaison avec les méthodes de pointe de régularisation pour CNN telles que Mixup, Cutout, CutMix, ManifoldMixup et Puzzle Mix. Mots clés: Apprentissage en profondeur, Réseau Neuronal Convolutif, Généralisation,Régularisation, Techniques de régularisation dépendantes et indépendantes des données, Robustesse aux attaques adverses. / Large capacity deep learning models are often prone to a high generalization gap when trained with a limited amount of labeled training data. And, in this case, very deep and wide networks have a tendency to memorize the samples, and therefore they might be vulnerable under a slight distribution shift at testing time. This problem yields poor generalization for data outside of the training data distribution. To overcome this issue some data-dependent and data-independent methods have been proposed. A recent class of successful methods to address this problem uses various ways to construct a new training sample by mixing a pair (or more) of training samples. In this thesis, we introduce PatchUp, a feature-space block-level data-dependent regularization that operates in the hidden space by masking out contiguous blocks of the feature map of a random pair of samples, and then either mixes (Soft PatchUp) or swaps (Hard PatchUp) these selected contiguous blocks. Our regularization method does not incur significant computational overhead for CNNs during training. Our approach improves the robustness of CNN models against the manifold intrusion problem that may occur in other state-of-the-art mixing approaches like Mixup and CutMix. Moreover, since we are mixing the contiguous block of features in the hidden space, which has more dimensions than the input space, we obtain more diverse samples for training towards different dimensions. Our experiments on CIFAR-10, CIFAR-100, SVHN, and Tiny-ImageNet datasets using ResNet architectures including PreActResnet18, PreActResnet34, WideResnet-28-10, ResNet101, and ResNet152 models show that PatchUp improves upon, or equals, the performance of current state-of-the-art regularizers for CNNs. We also show that PatchUp can provide a better generalization to affine transformations of samples and is more robust against adversarial attacks. PatchUp also helps a CNN model to produce a wider variety of features in the residual blocks compared to other state-of-the-art regularization methods for CNNs such as Mixup, Cutout, CutMix, ManifoldMixup, and Puzzle Mix. Key words: Deep Learning, Convolutional Neural Network, Generalization, Regular-ization, Data-dependent and Data-independent Regularization Techniques, Robustness to Adversarial Attacks.