Zentraler Gegenstand dieser Dissertation ist die Entwicklung von mathematischen Methoden zur Charakterisierung und Implementierung von optimalen Investmentstrategien eines Kleininvestors auf einem Finanzmarkt. Zur Behandlung dieser Probleme ziehen wir als Hauptwerkzeug Stochastische Rückwärts-Differenzialgleichungen (BSDEs) mit nicht-linearen Drifts heran. Diese Nicht-Lineariäten ordnen sie außerhalb der Standardklasse der Lipschitz-stetigen BSDEs ein und treten häufig in finanzmathematischen Kontrollproblemen auf. Wir charakterisieren das optimale Vermögen und die optimale Investmentstrategie eines Kleininvestors mit Hilfe einer sog. Stochastischen Vorwärts-Rückwärts-Differenzialgleichung (FBSDE), einem System bestehend aus einer stochastischen Vorwärtsgleichung, die vollständig gekoppelt ist an eine Rückwärtsgleichung. Die Festlegung bestimmter Nutzenfunktionen führt uns schließlich zu einer weiteren Klasse von nicht-standard BSDEs, die in unmittelbarem Zusammenhang zu dem sog. Ansatz der stochastischen partiellen Rückwärts-Differenzialgleichungen (BSPDEs) steht. Anschließend entwickeln wir eine Methode zur numerischen Behandlung von quadratischen BSDEs, die auf einem stochastischen Analogon der Cole-Hopf-Transformation basiert. Wir studieren weiterhin eine Klasse von BSDEs, deren Drifts explizite Pfadabhängigkiten aufweisen und leiten mehrere analytische Eigenschaften her. Schließlich studieren wir Dualdarstellungen für Optimalen Mehrfachstoppprobleme. Wir leiten Martingal-Dualdarstellungen her, die die Grundlage für die Entwicklung von Regressions-basierten Monte Carlo Simulationsalgorithmen bilden, die schnell und effektiv untere und obere Schranken berechnen. / This thesis elaborates on the wealth maximization problem of a small investor who invests in a financial market. Key tools for our studies come across in the form of several classes of BSDEs with particular non-linearities, casting them outside the standard class of Lipschitz continuous BSDEs. We first give a characterization of a small investor''s optimal wealth and its associated optimal strategy by means of a systems of coupled equations, a forward-backward stochastic differential equation (FBSDE) with non-Lipschitz coefficients, where the backward component is of quadratic growth. We then examine how specifying concrete utility functions give rise to another class of non-standard BSDEs. In this context, we also investigate the relationship to a modeling approach based on random fields techniques, known by now as the backward stochastic partial differential equations (BSPDEs) approach. We continue with the presentation of a numerical method for a special type of quadratic BSDEs. This method is based on a stochastic analogue to the Cole-Hopf transformation from PDE theory. We discuss its applicability to numerically solve indifference pricing problems for contingent claims in an incomplete market. We then proceed to BSDEs whose drifts explicitly incorporate path dependence. Several analytical properties for this type of non-standard BSDEs are derived. Finally, we devote our attention to the problem of a small investor who is equipped with several exercise rights that allow her to collect pre-specified cashflows. We solve this problem by casting it into the language of multiple optimal stopping and develop a martingale dual approach for characterizing the optimal possible outcome. Moreover, we develop regression based Monte Carlo algorithms which simulate efficiently lower and upper price bounds.

This master thesis describes the design of a track-and-hold (T&H) circuit with 10bit resolution, 80MS/s and 30MHz bandwidth. It is designed in a 0.18µm CMOS process with a supply voltage of 1.8 Volt. The circuit is supposed to work together with a 10bit pipelined analog to digital converter. A switched capacitor topology is used for the T&H circuit and the amplifier is a folded cascode OTA with regulated cascode. The switches used are of transmission gate type. The thesis presents the design decisions, design phase and the theory needed to understand the design decisions and the considerations in the design phase. The results are based on circuit level SPICE simulations in Cadence with foundry provided BSIM3 transistor models. They show that the circuit has 10bit resolution and 7.6mW power consumption, for the worst-case frequency of 30MHz. The requirements on the dynamic performance are all fulfilled, most of them with large margins.

<p>This master thesis describes the design of a track-and-hold (T&H) circuit with 10bit resolution, 80MS/s and 30MHz bandwidth. It is designed in a 0.18µm CMOS process with a supply voltage of 1.8 Volt. The circuit is supposed to work together with a 10bit pipelined analog to digital converter. </p><p>A switched capacitor topology is used for the T&H circuit and the amplifier is a folded cascode OTA with regulated cascode. The switches used are of transmission gate type. </p><p>The thesis presents the design decisions, design phase and the theory needed to understand the design decisions and the considerations in the design phase. </p><p>The results are based on circuit level SPICE simulations in Cadence with foundry provided BSIM3 transistor models. They show that the circuit has 10bit resolution and 7.6mW power consumption, for the worst-case frequency of 30MHz. The requirements on the dynamic performance are all fulfilled, most of them with large margins.</p>

Les composants de puissance grand gap présentent d'ores et déjà des caractéristiques statiques et dynamiques supérieures à leurs homologues en silicium. Mais ces composants d'un nouvel ordre s'accompagnent de différences susceptibles de modifier le fonctionnement de la cellule de commutation. Les travaux qui furent menés au cours de cette thèse se sont intéressés aux composants grand gap et à leur commande au sein d'un convertisseur de puissance synchrone robuste, haut rendement et haute densité de puissance. En particulier deux points critiques ont été identifiés et étudiés. Le premier est la grande sensibilité des composants grand gap aux composants parasites. Le second est l'absence de diode parasite interne entre le drain et la source de nombreux transistors grand gap. Pour répondre aux exigences de ces nouveaux composants et en tirer le meilleur profit, nous proposons des solutions innovantes, robustes, efficaces et directement intégrables aux circuits de commande. Des circuits de commande entièrement intégrés ont ainsi été conçus spécifiquement pour les composants grand gap. Ceux-ci permettent entre autres le contrôle précis des formes de commutation par l'adaptation de l'impédance de grille, et l'amélioration de l'efficacité énergétique et de la robustesse d'un convertisseur de puissance à base de composants grand sans diodes par une gestion dynamique et locale de temps morts très courts. / Wide band gap devices already demonstrate static and dynamic performances better than silicon transistors. Compared to conventional silicon devices these new wide band gap transistors have some different characteristics that may affect power converter operations. The work presented in this PhD manuscript deals with a specific gate drive circuit for a robust, high power density and high efficiency wide band gap devices-based power converter. Two critical points have been especially studied. The first point is the higher sensitivity of wide band gap transistors to parasitic components. The second point is the lack of parasitic body diode between drain and source of HEMT GaN and JFET SiC. In order to drive these new power devices in the best way we propose innovative, robust and efficient solutions. Fully integrated gate drive circuits have been specifically developed for wide band gap devices. An adaptive output impedance gate driver provides an accurate control of wide band gap device switching waveforms directly on its gate side. Another gate drive circuit improves efficiency and reliability of diode-less wide band gap devices-based power converters thanks to an auto-adaptive and local dead-time management.

The demand for flat glass is high and increasing significantly in the building industry as a direct result of architectural requirements for lightness, transparency and natural light. Current architectural trends require glass in curvilinear forms for smooth free-form façades. Two principal challenges arise from this: to cost-effectively produce the desired curvature and; to ensure its safe performance after exposure to ageing. The recent availability of high strength glass provides an opportunity to address the first challenge by developing cold bent glass. Cold bending involves the straining of relatively thin glass components, at ambient temperatures, and is a low energy and cost effective manner of creating curvilinear forms. However, cold bending is not yet widely established as a reliable method. The aim of this thesis is to develop the understanding of cold bent glass during the bending process and to evaluate its post-ageing performance. This thesis, firstly, investigates the mechanical response of monolithic glass plates during the cold bending process. The stability of cold bent glass is investigated experimentally by bending it in double curved anticlastic shapes. A parametric numerical analysis involves different boundary conditions, geometrical plate characteristics and bending parameters. The principal outcome is that a local instability, now termed cold bending distortion, occurs when certain displacement limits are exceeded and could degrade the optical quality of the glass. An evaluation procedure is also formulated to set limits and aid designers/manufacturers to predict the mechanical response and the optical quality of the glass. Cold bent glass is subjected to permanent bending stresses throughout its service life and therefore, its strength degradation after ageing needs to be quantified. Analytical, experimental and numerical investigations are undertaken in this thesis to identify the most effective method for estimating glass strength (evaluation of destructive tests, required number of specimens, statistical analysis methods and sub-critical crack growth). The limited availability of naturally aged toughened glass and the absence of a reliable ageing standard impede the evaluation of its aged performance. Therefore, a parametric experimental investigation of artificial ageing methods on glass is undertaken in this thesis. A procedure for the evaluation of the strength of aged glass is finally, formulated to allow the selection of artificial ageing parameters that correspond to a target level of erosion. The knowledge on artificial ageing and strength prediction acquired above is finally implemented on different types of glass to determine their strength after ageing and assess their safe use in cold bending / load bearing applications. The investigation showed that fully toughened glass has a superior performance to chemically toughened or annealed glass. Overall, the research presented in this thesis demonstrates that high quality cold bent toughened glass can be created when certain applied displacement limits are respected. These can be used as a safe, cost-effective and energy efficient replacement to the more conventional hot bent glass. However, cold bending / load bearing applications in which the stressed glass surface is exposed to ageing, require glass with a relatively high case depth such as fully toughened or bi-tempered glass.

Ultra low power wireless sensors and sensor systems are of increasing interest in a variety of applications ranging from structural health monitoring to industrial process control. Electrochemical batteries have thus far remained the primary energy sources for such systems despite the finite associated lifetimes imposed due to limitations associated with energy density. However, certain applications (such as implantable biomedical electronic devices and tire pressure sensors) require the operation of sensors and sensor systems over significant periods of time, where battery usage may be impractical and add cost due to the requirement for periodic re-charging and/or replacement. In order to address this challenge and extend the operational lifetime of wireless sensors, there has been an emerging research interest on harvesting ambient vibration energy. Vibration energy harvesting is a technology that generates electrical energy from ambient kinetic energy. Despite numerous research publications in this field over the past decade, low power density and variable ambient conditions remain as the key limitations of vibration energy harvesting. In terms of the piezoelectric transducers, the open-circuit voltage is usually low, which limits its power while extracted by a full-bridge rectifier. In terms of the interface circuits, most reported circuits are limited by the power efficiency, suitability to real-world vibration conditions and system volume due to large off-chip components required. The research reported in this thesis is focused on increasing power output of piezoelectric transducers and power extraction efficiency of interface circuits. There are five main chapters describing two new design topologies of piezoelectric transducers and three novel active interface circuits implemented with CMOS technology. In order to improve the power output of a piezoelectric transducer, a series connection configuration scheme is proposed, which splits the electrode of a harvester into multiple equal regions connected in series to inherently increase the open-circuit voltage generated by the harvester. This topology passively increases the rectified power while using a full-bridge rectifier. While most of piezoelectric transducers are designed with piezoelectric layers fully covered by electrodes, this thesis proposes a new electrode design topology, which maximizes the raw AC output power of a piezoelectric harvester by finding an optimal electrode coverage. In order to extract power from a piezoelectric harvester, three active interface circuits are proposed in this thesis. The first one improves the conventional SSHI (synchronized switch harvesting on inductor) by employing a startup circuitry to enable the system to start operating under much lower vibration excitation levels. The second one dynamically configures the connection of the two regions of a piezoelectric transducer to increase the operational range and output power under a variety of excitation levels. The third one is a novel SSH architecture which employs capacitors instead of inductors to perform synchronous voltage flip. This new architecture is named as SSHC (synchronized switch harvesting on capacitors) to distinguish from SSHI rectifiers and indicate its inductorless architecture.

Diese Diplomarbeit beschäftigt sich mit der geometrischen Auslegung und Regelung einer vollaktuierten Drohne, die als fliegende Arbeitsplattform für einen Manipulator dienen soll. Dabei werden ausgehend von der geometrischen Beschreibung einer allgemeinen, symmetrischen Drohne mit sechs Rotoren Methoden entwickelt, die den anforderungsbezogenen Entwurf der Geometrie einer vollaktuierten Drohne ermöglichen. Darüber hinaus werden prinzipielle Einflussmechanismen einzelner Geometrieparameter auf die durch die Drohne erzeugbaren Kräfte und Momente aufgezeigt. Zur Charakterisierung des Raums aller erzeugbaren Lasten wird dieser auf sogenannte Stützvektoren reduziert. Als Stützvektoren dienen dabei die für den Schwebeflug nötige Schubkraft, die garantierte Mindestkraft in horizontaler Richtung und das garantierte Mindestmoment um eine beliebige Achse, zu deren Berechnung zusätzlich analytische Formeln hergeleitet werden. Aufbauend auf die Beschreibung durch Stützvektoren wird die Formuliernung von Metriken vorgestellt, die die Bewertung einer Drohnengeometrie durch eine einzige skalare Maßzahl ermöglichen, wodurch die je nach Anwendung optimale Drohnengeometrie ermittelt werden kann. Zur Regelung des Systems aus Drohne und Manipulator wurde ein Regelungskonzept entwickelt, welches durch eine Entkopplung der Bewegungsgleichungen eine virtuelle Verschiebung des Schwerpunkts in das Drohnenzentrum realisiert und so eine präzise Regelung unabhängig von der tatsächlichen Schwerpunktlage ermöglicht. / This thesis’ subject is the geometrical design and control of a fully actuated drone, intended to be used as a flying operating-platform for a manipulator. Starting with the general geometrical description of a symmetric drone with six rotors, methods for the application specific design of a fully actuated drone are developed. Furthermore general influencing principles of geometric parameters on the forces and torques that can be generated by the drone, are pointed out. To characterize the drone's wrench-space, it is reduced to so called support vectors, which are given by the hovering thrust, the minimum guaranteed force in a horizontal direction and the minimum guaranteed torque in any direction. Additionally, analytic formulas are derived for the mentioned support vectors. Based on the description by the support vectors, a formulation of metrics is introduced, to enable the assessment of a specific drone geometry by a single scalar measure, to determine the ideal drone geometry for a specific application. Targeting the issue of controlling the flight system, consisting of the drone and the manipulator, a concept is developed that realizes a virtual dissplacement of the center of mass by decoupling the equations of motion and therby facilitates a precise control, independent of the actual location of the system's center of mass.

Over the last decades, there has been considerable attention in the accurate mathematical modeling and numerical simulations of free surface wave propagation in near-shore environments. A physical correct description of the large scale phenomena, which take place in the shallow water region, must account for strong nonlinear and dispersive effects, along with the interaction with complex topographies. First, a study on the behavior in nonlinear regime of different Boussinesq-type models is proposed, showing the advantage of using fully-nonlinear models with respect to weakly-nonlinear and weakly dispersive models (commonly employed). Secondly, a new flexible strategy for solving the fully-nonlinear and weakly-dispersive Green-Naghdi equations is presented, which allows to enhance an existing shallow water code by simply adding an algebraic term to the momentum balance and is particularly adapted for the use of hybrid techniques for wave breaking. Moreover, the first discretization of the Green-Naghdi equations on unstructured meshes is proposed via hybrid finite volume/ finite element schemes. Finally, the models and the methods developed in the thesis are deployed to study the physical problem of bore formation in convergent alluvial estuary, providing the first characterization of natural estuaries in terms of bore inception. / Ces dernières décennies, une attention particulière a été portée sur la modélisation mathématique et la simulation numérique de la propagation de vagues en environnements côtiers. Une description physiquement correcte des phénomènes à grande échelle, qui apparaissent dans les régions d'eau peu profonde, doit prendre en compte de forts effets non-linéaires et dispersifs, ainsi que l'interaction avec des bathymétries complexes. Dans un premier temps, une étude du comportement en régime non linéaire de différents modèles de type Boussinesq est proposée, démontrant l'avantage d'utiliser des modèles fortement non-linéaires par rapport à des modèles faiblement non-linéaires et faiblement dispersifs (couramment utilisés). Ensuite, une nouvelle approche flexible pour résoudre les équations fortement non-linéaires et faiblement dispersives de Green-Naghdi est présentée. Cette stratégie permet d'améliorer un code "shallow water" existant par le simple ajout d'un terme algébrique dans l'équation du moment et est particulièrement adapté à l'utilisation de techniques hybrides pour le déferlement des vagues. De plus, la première discrétisation des équations de Green-Naghdi sur maillage non structuré est proposée via des schémas hybrides Volume Fini/Élément Fini. Finalement, les modèles et méthodes développés dans la thèse sont appliqués à l'étude du problème physique de la formation du mascaret dans des estuaires convergents et alluviaux. Cela a amené à la première caractérisation d'estuaire naturel en terme d'apparition de mascaret.

Under hösten 2019 har Vattenfall Research &amp; Development byggt en experimentell jordfyllningsdamm i Älvkarleby med dimensionerna 20x15x4 meter. Delar av experimentdammen är konventionellt konstruerade och har installerats med geoteknisk utrustning som utgörs av bland annat inklinometrar och portrycksgivare. Andra delar av experimentdammen har byggts in med defekter som ska representera åldersrelaterade skador eller utförandefel vid konstruktion. Experimentdammen ger möjlighet att under realistiska och kontrollerade förhållanden studera det mekaniska beteendet i samband med fyllning av vatten och vidare drift med hjälp av den geotekniska instrumenteringen samt med stöd av numerisk modellering. I detta examensarbete, som utgör en del av Luleå tekniska universitets forskningsprojekt mot experimentdammen, har experimentdammens beteende i form av deformationer och portryck studerats under uppfyllnad och drift fram till sommaren 2021. Detta har utförts genom simuleringar i det finita elementprogrammet PLAXIS 2D 2019 för en tvärsektion av experimentdammen i plant-deformationstillstånd. Mätpunkterna i modellen har baserats på faktisk placering av den geotekniska instrumenteringen. Den finita elementmodellen av experimentdammen har konstruerats och fyllts med vatten enligt dokumentation från fält. En flödes-deformationsanalys, med den konstitutiva modellen Hardening Soil och den hydrauliska modellen van Genuchten, har tillämpats för att modellera den simultana utvecklingen av portryck och deformationer under uppfyllnad. Materialparametervärden för den finita elementmodelleringen har erhållits från Vattenfall R&amp;D, relevant litteratur och från fält- och laboratorieförsök. I fält har vattenvolymeterförsök utförts på tätkärnan och i laboratoriemiljö har modifierad proctorpackning, dränerade konventionella triaxialförsök, permeabilitetsförsök och övertryckskapillarimeterförsök utförts på tätkärnans material. Resultatet från övertryckskapillarimeterförsök har anpassats mot den hydrauliska modellen van Genuchten för att uppskatta en vattenbindningskurva som beskriver det icke-linjära förhållandet mellan jordens vatteninnehåll och porundertryck, det vill säga det omättade förhållandet. Vattenbindningskurvor för övriga materialzoner har uppskattats baserat på litteratur. Verktyget PLAXIS SoilTest har använts för att optimera materialparametervärden för tätkärnan mot resultat från utförda triaxialförsök. Materialparametrarna E50ref, Eoedref, Eurref, m, c, och ϕ har optimerats fram till brott i triaxial belastning. En känslighetsanalys har utförts för reduktion av filterzonernas och stödfyllningens styvhetsmoduler och deras inverkan på horisontella deformationer i dammkroppen under uppfyllnad. Känslighetsanalysen indikerar att finfiltrets styvhetsmoduler har störst inverkan och grovfiltrets styvhetsmoduler har minst inverkan på de horisontella deformationerna. Studiens resultat visar att magnituden av horisontella och vertikala deformationer kommer vara som störst i den övre delen av dammkroppen och uppgår där till 3,5 respektive 4,0 mm. Dammkroppens huvudsakliga rörelse kommer vara i nedströms riktning och det observerades hur en kontaktzon mellan uppströms filterzon och tätkärnan utgör en gräns för riktning av deformationer. Faktiskt uppmätta rörelser i installerade inklinometrar kunde inte jämföras mot deformationer i den finita elementmodellen eftersom författarens tolkning indikerar på att botten av inklinometrarna har rört på sig, och mätpunkterna i botten av modellen är fixerade. Modellen visar hur en fördröjd utveckling av vattenmättnad sker genom tätkärnan, där uppströms sida av tätkärnan reagerar snabbare på förändringar i vattennivå jämfört med nedströms sida av tätkärnan som uppvisar en fördröjd respons. Vid en sänkning av vattennivån observerades hur tätkärnan håller kvar vatten ovan portryckslinjen medan de grövre materialen dränerar i takt med vattennivåns sänkning. Utvecklingen av de simulerade portrycken i modellen under uppfyllnad och drift överensstämmer bra med de uppmätta portrycken i experimentdammen, när portrycken är positiva. Det observeras hur den finita elementmodellen överskattar negativa portryck (porundertryck). Portrycken i modellen når ett stadigt tillstånd ungefär 115 dagar efter att fyllningen av vatten påbörjats. Den finita elementmodellen lyckas att återge det teoretiska beteendet av jordfyllningsdammar under fyllning i form av huvudsakliga riktningar av deformationer och utveckling av vattenmättnad i tätkärnan. Denna studie bidrar till en djupare förståelse för experimentdammens, och i allmänhet jordfyllningsdammars, mekaniska beteende under uppfyllnad. Resultaten från den finita elementmodellen kan ur ett dammsäkerhetsperspektiv användas för erhålla indikationer på utvecklingen av deformationer, portryck och vattenmättnadsgrad i jordfyllningsdammar under uppfyllnad, och även under en tillfällig sänkning av vattennivån under den första fyllningen. Studien ger också indikationer på vilka materialparametrar som är viktiga vid numerisk modellering av mekaniskt beteende i jordfyllningsdammar. / During the autumn of 2019, Vattenfall Research &amp; Development constructed an experimental embankment dam in Älvkarleby with the dimensions 20x15x4 metres. Parts of the experimental dam are conventionally constructed and have been equipped with geotechnical instrumentation which consist of, among other things, inclinometers and pore pressure transducers. Other parts of the experimental dam have built in defects to represent age-related damages or execution errors during construction. The experimental embankment dam provides the opportunity to, under realistic and controlled conditions, study the mechanical behaviour during filling of water and operation by means of the geotechnical instrumentation and the use of numerical modelling. In this master’s thesis, which forms part of Luleå University of Technology’s research project towards the experimental dam, the behaviour of the experimental dam in terms of deformations and pore pressures have been studied during filling and operation until the summer of 2021. This has been performed by simulations in the finite element program PLAXIS 2D 2019 for a cross section of the experimental dam under plane-strain conditions. Measuring points in the model have been based on the actual location of the geotechnical instrumentation. The finite element model of the experimental dam has been constructed and filled according to documentation from field. A fully-coupled flow deformation analysis, with the constitutive model Hardening Soil and hydraulic model van Genuchten, has been utilised to model the simultaneous development of pore pressure and deformations during filling. Values of material parameters for the finite element modelling have been received from Vattenfall R&amp;D, relevant literature and from field- and laboratory tests. In the field, balloon tests have been performed on the core material. In laboratory environment, modified proctor compaction tests, drained conventional triaxial tests, permeability tests and pressure plate tests have been performed on the core material. Results from the pressure plate tests have been adapted to the hydraulic model van Genuchten to estimate a soil-water characteristic curve in order to describe the non-linear relation between the water content and suction in the soil, i.e. unsaturated conditions. Soil-water characteristic curves for the other material zones have been estimated based on literature. The tool PLAXIS SoilTest has been used to optimise material parameter values of the core against the results from conducted triaxial tests. The material parameters E50ref, Eoedref, Eurref, m, c, and ϕ have been optimised until failure in triaxial loading. A sensitivity analysis has been carried out, by reducing stiffness moduli of the filter zones and the shoulder material, to investigate the influence on horizontal deformations in the dam body during filling. The sensitivity analysis indicates that the stiffness moduli of the fine filter have the largest impact and the stiffness moduli of the coarse filter have the least impact on the horizontal deformations. The results of the study show that the magnitude of horizontal and vertical deformations will be largest in the upper part of the dam body and amounts to 3.5 and 4.0 mm, respectively. The main movement of the dam body will be in the downstream direction and it was observed how a contact zone between the upstream filter zone and the core forms a boundary for direction of deformations. Actual measured movements in the installed inclinometers could not be compared to deformations in the finite element model because the author’s interpretation indicates that the bottom of the inclinometers have moved, and the measuring points at the bottom of the model are fixed. The model shows how a delayed development of saturation occur through the core, where the upstream side of the core responds more quickly to changes in water level compared with the downstream side of the core which show a delayed response. At a lowering of the water level, it was observed how the core retains water above the phreatic line while the coarser materials drain as the water level decreases. Development of the simulated pore pressures in the model during filling and operation corresponds well with the measured pore pressures in the experimental dam, when the pore pressures are positive. It is observed how the finite element model overestimates negative pore pressures (suction). The pore pressures in the model reaches a steady state approximately 115 days after filling of water started. The finite element model succeeds in reproducing the theoretical behaviour of embankment dams during filling in terms of main directions of deformations and development of saturation in the core. This study contributes to a deeper understanding of the experimental dam, and in general mechanical behaviour of embankment dams during filling. The results from the finite element model can be used from a dam safety perspective to obtain indications on the development of deformations, pore pressures and degree of saturation in embankment dams during filling, and also for a temporary lowering of the water level during the first filling. The study also provides indications of which material parameters that are of importance in numerical modelling of mechanical behaviour in embankment dams.

Cílem této diplomové práce je automatizace detekce vad ve vláknitých materiálech. Firma SILON se již přes padesát let zabývá výrobou jemné vaty z recyklovaných PET lahví. Tato vata se následně používá ve stavebnictví, automobilovém průmyslu, ale nejčastěji v dámských hygienických potřebách a dětských plenách. Cílem firmy je produkovat co nejkvalitnější výrobek a proto je každá dávka testována v laboratoři s několika přísnými kritérii. Jednám z testů je i množství vadných vláken, jako jsou zacuchané smotky vláken, nebo nevydloužená vlákna, která jsou tvrdá a snadno se lámou. Navrhovaný systém sestává ze snímací lavice fungující jako scanner, která nasnímá vzorek vláken, který byl vložen mezi dvě skleněné desky. Byla provedena série testů s různým osvětlením, která ověřovala vlastnosti Rhodaminu, který se používá právě na rozlišení defektů od ostatních vláken. Tyto defekty mají zpravidla jinou molekulární strukturu, na kterou se barvivo chytá lépe. Protože je Rhodamin fluorescenční barvivo, je možné ho například pod UV světlem snáze rozeznat. Tento postup je využíván při manuální detekci. Při snímání kamerou je možno si vypomoci filtrem na kameře, který odfiltruje excitační světlo a propustí pouze světlo vyzářené Rhodaminem. Součástí výroby skeneru byla i tvorba ovládacího programu. Byla vytvořena vlastní knihovna pro ovládání motoru a byla upravena knihovna pro kameru. Oba systém pak bylo možno ovládat pomocí jednotného GUI, které zajišťovalo pořizování snímku celé desky. Pomocí skeneru byla nasnímána řada snímků, které bylo třeba anotovat, aby bylo možné naučit počítač rozlišovat defekty. Anotace proběhla na pixelové úrovni; každý defekt byl označen v grafickém editoru ve speciální vrstvě. Pro rozlišování byla použita umělá neuronová síť, která funguje na principu konvolucí. Tento typ sítě je navíc plně konvoluční, takže výstupem sítě je obraz, který by měl označit na tom původním vadné pixely. Výsledky naučené sítě jsou v práci prezentovány a diskutovány. Síť byla schopna se naučit rozeznávat většinu defektů a spolehlivě je umí rozeznat a segmentovat. Potíže má v současné době s detekcí rozmazaných defektů na krajích zorného pole a s defekty, jejichž hranice není tolik zřetelná na vstupních obrazech. Nutno zmínit, že zákazník má zájem o kompletní řešení scanneru i s detekčním softwarem a vývoj tohoto zařízení bude pokračovat i po závěru této diplomové práce.