Thesis (MSc)--Stellenbosch University, 2015. / ENGLISH ABSTRACT: This study has started as part of a research project at Stellenbosch University (SU) that aims at building a team of soccer-playing robots for the RoboCup Small Size League (SSL). In the RoboCup SSL the Decision- Making Module (DMM) plays an important role for it makes all decisions for the robots in the team. This research focuses on the development of some parts of the DMM for the team at SU. A literature study showed that the DMM is typically developed in a hierarchical structure where basic soccer skills form the fundamental building blocks and high-level team behaviours are implemented using these basic soccer skills. The literature study also revealed that strategies in the DMM are usually developed using a hand-coded approach in the RoboCup SSL domain, i.e., a specific and fixed strategy is coded, while in other leagues a Machine Learning (ML) approach, Reinforcement Learning (RL) in particular, is widely used. This led to the following research objective of this thesis, namely to develop basic soccer skills using RL for the RoboCup Small Size League. A second objective of this research is to develop a simulation environment to facilitate the development of the DMM. A high-level simulator was developed and validated as a result. The temporal-difference value iteration algorithm with state-value functions was used for RL, along with a Multi-Layer Perceptron (MLP) as a function approximator. Two types of important soccer skills, namely shooting skills and passing skills were developed using the RL and MLP combination. Nine experiments were conducted to develop and evaluate these skills in various playing situations. The results showed that the learning was very effective, as the learning agent executed the shooting and passing tasks satisfactorily, and further refinement is thus possible. In conclusion, RL combined with MLP was successfully applied in this research to develop two important basic soccer skills for robots in the RoboCup SSL. These form a solid foundation for the development of a complete DMM along with the simulation environment established in this research. / AFRIKAANSE OPSOMMING: Hierdie studie het ontstaan as deel van 'n navorsingsprojek by Stellenbosch Universiteit wat daarop gemik was om 'n span sokkerrobotte vir die RoboCup Small Size League (SSL) te ontwikkel. Die besluitnemingsmodule (BM) speel 'n belangrike rol in die RoboCup SSL, aangesien dit besluite vir die robotte in die span maak. Hierdie navorsing fokus op ontwikkeling van enkele komponente van die BM vir die span by SU. 'n Literatuurstudie het getoon dat die BM tipies ontwikkel word volgens 'n hiërargiese struktuur waarin basiese sokkervaardighede die fundamentele boublokke vorm en hoëvlak spangedrag word dan gerealiseer deur hierdie basiese vaardighede te gebruik. Die literatuur het ook getoon dat strategieë in die BM van die RoboCup SSL domein gewoonlik ontwikkel word deur 'n hand-gekodeerde benadering, dit wil s^e, 'n baie spesifieke en vaste strategie word gekodeer, terwyl masjienleer (ML) en versterkingsleer (VL) wyd in ander ligas gebruik word. Dit het gelei tot die navorsingsdoelwit in hierdie tesis, naamlik om basiese sokkervaardighede vir robotte in die RoboCup SSL te ontwikkel. 'n Tweede doelwit was om 'n simulasie-omgewing te ontwikkel wat weer die ontwikkeling van die BM sou fasiliteer. Hierdie simulator is suksesvol ontwikkel en gevalideer. Die tydwaarde-verskil iterariewe algoritme met toestandwaarde-funksies is gebruik vir VL saam met 'n multi-laag perseptron (MLP) vir funksiebenaderings. Twee belangrike sokkervaardighede, naamlik doelskop- en aangeevaardighede is met hierdie kombinasie van VL en MLP ontwikkel. Nege eksperimente is uitgevoer om hierdie vaardighede in verskillende speelsituasies te ontwikkel en te evalueer. Volgens die resultate was die leerproses baie effektief, aangesien die leer-agent die doelskiet- en aangeetake bevredigend uitgevoer het, en verdere verfyning is dus moontlik. Die gevolgtrekking is dat VL gekombineer met MLP suksesvol toegepas is in hierdie navorsingswerk om twee belangrike, basiese sokkervaardighede vir robotte in die RoboCup SSL te ontwikkel. Dit vorm 'n sterk fondament vir die ontwikkeling van 'n volledige BM tesame met die simulasie-omgewing wat in hierdie werk daargestel is.

La technologie SIW, introduite dans les années 2000, suscite aujourd’hui un très vif intérêt pour la conception de circuits micro-ondes compacts, intégrés, faible coût et blindés par nature. Cependant, les guides d’onde métalliques, qui offrent de bien meilleures performances en termes de pertes d’insertion et de tenue en puissance, malgré un coût bien plus important, sont encore incontournables pour de nombreuses applications millimétriques. Afin de proposer une alternative intégrée et faible coût au guide d’onde rectangulaire, et de permettre une large exploitation du spectre millimétrique, cette thèse propose une nouvelle structure SIW appelée SIW creux. Cette nouvelle structure a été étudiée théoriquement et expérimentalement. Aux fréquences millimétriques, comparativement au SIW, le SIW creux offre des pertes d’insertion trois fois plus faible ainsi qu’une tenue en puissance moyenne quatre fois plus importante. De nombreux dispositifs passifs SIW creux ont été conçus en prenant avantage du procédé de circuit imprimé multicouche mis en œuvre. Des coupleurs, déphaseurs, diviseurs de puissance, antennes et filtres ont été réalisés basés sur la technologie introduite. Leurs performances sont théoriquement et expérimentalement comparées avec leur contrepartie SIW afin de démontrer les avantages de la nouvelle technologie proposée. / The substrate integrated waveguide (SIW) technology, introduced in the early 2000s, has presently trigged a huge interest from academia to industry with the focus on the design and development of low-loss, compact, integrated, self-packaged and low-cost microwave and millimeter-wave circuits, antennas and systems. However, the classical metallic waveguide technology, which offers better performances such as lower insertion loss and higher power handling, has still been used in the design of microwave and millimeter-wave systems, despite its higher cost and bulky structure. To offer a highly integrated, further loss-reduced, low-cost alternative to the conventional waveguide and also to allow a wide-spread use of the millimeter-wave spectrum, this thesis research introduces a new SIW structure called Air-Filled SIW (AFSIW). This new structure has been theoretically and experimentally studied in details with a substantial amount of results. At millimeter wave frequencies, compared to the SIW topologies, the proposed AFSIW scheme exhibits a substantially lower insertion loss (three times, for example) and a much higher average power handling capability (four times, for example). Numerous AFSIW passive components have been investigated designed and demonstrated, which take advantages of the well-established multilayer printed circuit board (PCB) fabrication process. Couplers, phase shifters, power dividers, antennas and filters have been modeled, designed, prototyped and measured based on the introduced technology. Their performances have theoretically and experimentally been compared with their SIW counterparts to demonstrate and validate the benefits of the proposed technology.

Adaptive coatings are an important development in tribology. These coatings widen the range at which solid lubricants are useful in various environments. In this paper, coatings founded on molybdenum nitride are studied, with a focus on thermal cycling. These coatings were fabricated by unbalanced magnetron sputtering and characterized with techniques including x-ray diffraction (XRD), scanning electron microscopy (SEM), Raman spectroscopy, energy dispersive x-ray spectroscopy (EDS), and pin-on-disk tribometer. The results of two sets of coatings are reported. The first set of coatings is a nanocomposite of Mo2N/MoS2/Me (Me = Ag, Au, Cu). The second is a complex multi-layer system of Mo2N/Ag and a diffusion barrier of TiN which has been etched, then filled and coated with a layer of MoS2. After heating, these compounds produced silver molybdates. The Mo2N/MoS2/Ag nanocomposite shows promise with a 0.02 coefficient of friction at room temperature, while the multi-layer system eventually equilibrated at approximately 0.6. At high temperatures, again the nanocomposite was better, producing a 0.25 frictional coefficient compared to a 0.3 from the multilayer system. These results provide insight into what is needed to achieve thermal cycling.

Dans ce travail de thèse, le comportement des composites stratifiés est étudié en utilisant des développements d'ordres élevés ou de type layer wise. Un modèle layer wise LS1 (et l'élément fini à huit nœuds qui lui est associé) a été développé en premier lieu pour l'analyse spécifique des contraintes entre les couches pour des problèmes de bords libres ou bien pour l'étude des liaisons entre les couches. Il est fondé sur un modèle de type layer wise qui considère le stratifié comme une superposition de plaques de Reissner couplées les unes aux autres par des efforts d'interfaces. Cet élément a 5n degrés de liberté par noeud (n étant le nombre de couches) et donne une prédiction des contraintes d'interface. Les travaux précédents se concentraient sur la précision et l'utilisation de l'estimation de ces contraintes d'interface. Mais le modèle permet aussi de calculer les déplacements et les déflexions ainsi que d'extraire les contraintes dans l'épaisseur avec ou sans post-traitement. Dans ce mémoire, la précision et la validité du calcul des déplacements et des contraintes ont été établies à partir d'un nombre important de benchmarks en flexion de composites et des plaques sandwich. Le but de cette étude est aussi de mieux positionner et de promouvoir cette approche en contraintes issue des travaux de Pagano peu utilisés et mal référencés. Ce modèle est ensuite développé et implémenté dans le code de calcul par éléments finis Abaqus via la routine utilisateur User-Element (UEL). Pour démontrer l'efficacité du modèle et du code de calcul, des exemples qui comprennent des plaques isotropes et anisotropes sont traités / In this document, the behaviour of laminated composite is investigated using several high order or layer wise finite element calculations. A layer wise model and its devoted finite element on C° eight-node layer wise finite element have been originally and specifically developed for an interlaminar stresses analysis in a free edge problem or for a bonding study. This model is the core of the present comparisons. It is based on a typical layer wise model that considers the laminate by a superposition of Reissner plates coupled by interfacial stresses. This element consists of 5n degrees of freedom per node (n is the layer number) and is able to predict interlaminar stresses. These out-of-plane stresses are deduced directly from constitutive equations without post-processing works. The previous papers dealt with the accuracy of these estimated interface stresses. However, the model not only permits to compute displacements and usual deflections but also to compute stresses through the layer thicknesses with or without postprocess. In the present paper, the accuracy and the validity of these computed displacements and stresses have also been established on a large number of important benchmarks of bending examples for composites and sandwich plates. The aim of this research is also to better positioning and promoting of this stress approach deriving from the works of Pagano, which was not generally studied in this way. This model has also been implemented into the ABAQUS finite-element program through a User-Element subroutine interface. To demonstrate the effectiveness of the model and the finite element code, examples which include isotropic and anisotropic plates, the problem of the perforated plate are calculated

In this research, a novel machine learning method is designed and applied to Mine Countermeasure Missions. Similarly to some kernel methods, the proposed approach seeks to compute a linear model from another higher dimensional feature space. However, no kernel is used and the feature mapping is explicit. Computation can be done directly in the accessible feature space. In the proposed approach, the feature projection is implemented by constructing a large hidden layer, which differs from traditional belief that Multi-Layer Perceptron is usually funnel-shaped and the hidden layer is used as feature extractor. The proposed approach is a general method that can be applied to various problems. It is able to improve the performance of the neural network based methods and the learning speed of support vector machine. The classification speed of the proposed approach is also faster than that of kernel machines on the mine countermeasure mission task.

A multilayered system is a good model for many optical systems. An optical coherence tomography (OCT) system can provide fundamental information about the refractive index distribution of the sample and enables images to be corrected for geometric accuracy, but this requires the separation of refractive index and physical thickness for each layer from the measured optical paths. In this thesis a novel approach for simultaneous extraction of index and thickness of multi layer systems and the last medium index of refraction in a single experiment using only the object’s spectral response available by any Fourier Domain OCT system without using any additional outside measurements is introduced. The method is based on a novel matrix equation that uses the reflected spectrum from the object and the measured optical thickness. In the presence of slight error of measuring optical thickness, the parameters extraction has wavenumber dependency. A novel method is used to select the suitable set of spectral components that reduces the extracted parameters error. The parameters extraction method is followed by a fitting process for optimized results. The method works the best for low contrast index distribution even in the presence of relatively large optical thickness measurement error.

In today’s electronics, transistors are the main building blocks of the vast majority of electronic devices and integrated circuits. Types of transistors vary depending on the device structure and operation principle. Metal-oxide-based thin film transistors (MO TFTs), in particular, are an emerging technology that has a promising future in many applications, such as large-area display and wearable electronics. It exhibits unique features that make it superior to the existing Si-based technology, such as optical transparency and mechanical flexibility. However, some technical challenges in MO TFTs limit their emplyoment in today’s applications, such as low carrier mobility and high processing temperature. Solution-processed MO TFT based on spray pyrolysis combined with a carefully engineered TFT structure offers a dramatically enhance carrier mobility at low processing temperature. In this work, we are utilizing spray pyrolysis to grow In2O3 and ZnO based TFTs at low processing temperature. The structural effects of the channel layer on the electrical performance is investigated in two parts. The first part highlights the impact of thickness of the channel layer on the device performance of both In2O3 and ZnO, while the second part explores In2O3/ZnO heterojunction-based active layer. The results showed that increasing the channel thickness of both In2O3 and ZnO based TFTs enhanced the carrier mobility due to a reduced surface-roughness scattering effect. In addition, evidence showed that the electron transport mechanism in In2O3/ZnO heterojunction transitioned from trap-limited conduction (TLC) to percolation conduction (PC) process. Thanks to the existence of a 2D-confined electron sheet at the atomically sharp In2O3/ZnO heterointerface, the electron mobility was dramatically enhanced.

Efficient usage of file systems poses a major challenge for highly scalable parallel applications. The performance of even the most sophisticated I/O subsystems lags behind the compute capabilities of current processors. To improve the utilization of I/O subsystems, several libraries, such as HDF5, facilitate the implementation of parallel I/O operations. These libraries abstract from low-level I/O interfaces (for instance, POSIX I/O) and may internally interact with additional I/O libraries. While improving usability, I/O libraries also add complexity and impede the analysis and optimization of application I/O performance. This thesis proposes a methodology to investigate application I/O behavior in detail. In contrast to existing approaches, this methodology captures I/O activities on multiple layers of the I/O software stack, correlates these activities across all layers explicitly, and identifies interactions between multiple layers of the I/O software stack. This allows users to identify inefficiencies at individual layers of the I/O software stack as well as to detect possible conflicts in the interplay between these layers. Therefor, a monitoring infrastructure observes an application and records information about I/O activities of the application during its execution. This work describes options to monitor applications and generate event logs reflecting their behavior. Additionally, it introduces concepts to store information about I/O activities in event logs that preserve hierarchical relations between I/O operations across all layers of the I/O software stack. In combination with the introduced methodology for multi-layer I/O performance analysis, this work provides the foundation for application I/O tuning by exposing patterns in the usage of I/O routines. This contribution includes the definition of I/O access patterns observable in the event logs of parallel scientific applications. These access patterns originate either directly from the application or from utilized I/O libraries. The introduced patterns reflect inefficiencies in the usage of I/O routines or reveal optimization strategies for I/O accesses. Software developers can use these patterns as a guideline for performance analysis to investigate the I/O behavior of their applications and verify the effectiveness of internal optimizations applied by high-level I/O libraries. After focusing on the analysis of individual applications, this work widens the scope to investigations of coordinated sequences of applications by introducing a top-down approach for performance analysis of entire scientific workflows. The approach provides summarized performance metrics covering different workflow perspectives, from general overview to individual jobs and their job steps. These summaries allow users to identify inefficiencies and determine the responsible job steps. In addition, the approach utilizes the methodology for performance analysis of applications using multi-layer I/O to record detailed performance data about job steps, enabling a fine-grained analysis of the associated execution to exactly pinpoint performance issues. The introduced top-down performance analysis methodology presents a powerful tool for comprehensive performance analysis of complex workflows. On top of their theoretical formulation, this thesis provides implementations of all proposed methodologies. For this purpose, an established performance monitoring infrastructure is enhanced by features to record I/O activities. These contributions complement existing functionality and provide a holistic performance analysis for parallel scientific applications covering computation, communication, and I/O operations. Evaluations with synthetic case studies, benchmarks, and real-world applications demonstrate the effectiveness of the proposed methodologies. The results of this work are distributed as open-source software. For instance, the measurement infrastructure including improvements introduced in this thesis is available for download and used in computing centers world-wide. Furthermore, research projects already employ the outcomes of this work.

Classification tasks are an integral part of science, industry, medicine, and business; being such a pervasive technique, its smallest improvement is valuable. Artificial Neural Network (ANN) is one of the strongest techniques used in many disciplines for classification. The ANN technique suffers from drawbacks such as intransparency in spite of its high prediction power. In this dissertation, motivated by learning styles in human brains, ANN’s shortcomings are assuaged and its learning power is improved. Self-Organizing Map (SOM), an ANN variation which has strong unsupervised power, and Feedforward ANN, traditionally used for classification tasks, are hybridized to solidify their benefits and help remove their limitations. These benefits are in two directions: enhancing ANN’s learning power, and improving decision-making. First, the proposed method, named Self-Organizing Error-Driven (SOED) Artificial Neural Network (ANN), shows significant improvements in comparison with usual ANNs. We show SOED is a more accurate, more reliable, and more transparent technique through experimentation with five famous benchmark datasets. Second, the hybridization creates space for inclusion of decision-making goals at the level of ANN’s learning. This gives the classifier the opportunity to handle the inconclusiveness of the data smarter and in the direction of decision-making goals. Through three case studies, naming 1) churn decision analytics, 2) breast cancer diagnosis, and 3) quality control decision making through thermal monitoring of additive manufacturing processes, this novel and cost-sensitive aspect of SOED has been explored and lead to much quantified improvement in decision-making.

京都大学 / 新制・課程博士 / 博士(理学) / 甲第24390号 / 理博第4889号 / 新制||理||1699(附属図書館) / 京都大学大学院理学研究科数学・数理解析専攻 / (主査)教授 大木谷 耕司, 教授 並河 良典, 准教授 竹広 真一 / 学位規則第4条第1項該当 / Doctor of Science / Kyoto University / DFAM