Die vorliegende Arbeit befasst sich mit einem detaillierten Vergleich von Mehrpunkt-Schaltungstopologien mit zentralem Gleichspannungszwischenkreis für den Einsatz in Mittelspannungsanwendungen. Im Rahmen dieser Untersuchungen wird die 3-Level Neutral Point Clamped Spannungswechselrichter Schaltungstopologie (3L-NPC VSC) sowohl mit Multilevel Flying Capacitor (FLC) als auch mit Multilevel Stacked Multicell (SMC) Schaltungstopologien verglichen, wobei unter Verwendung von aktuell verfügbaren IGBT-Modulen Stromrichterausgangsspannungen von 2.3 kV, 4.16 kV und 6.6 kV betrachtet werden. Neben der grundlegenden Funktionsweise wird die Auslegung der aktiven Leistungshalbleiter und der passiven Energiespeicher (Zwischenkreiskondensatoren, Flying Capacitors) für die untersuchten Stromrichtertopologien dargestellt. Unter Berücksichtigung verschiedener Modulationsverfahren und Schaltfrequenzen werden Kennwerte für den Oberschwingungsgehalt in der Ausgangsspannung und dem Ausgangsstrom vergleichend evaluiert. Die installierte Schalterleistungen, die Halbleiterausnutzungsfaktoren, die Stromrichterverlustleistungen sowie die Verlustleistungsverteilungen werden für die betrachteten Stromrichtertopologien detailliert gegenübergestellt und bewertet.:Inhaltsverzeichnis Liste der Variablen i Liste der Abkürzungen v 1 Einleitung 1 2 Überblick von Mittelspannungsstromrichtertopologien und Leistungshalbleitern 3 2.1 Mittelspannungsumrichtertopologien 3 2.2 Leistungshalbleiter 8 3 Aufbau und Funktion von Mittelspannungsstromrichtertopologien 10 3.1 Neutral Point Clamped Stromrichter (NPC) 10 3.1.1 3-Level Neutral Point Clamped Stromrichter (3L-NPC) 10 3.1.2 Mehrstufige NPC-Umrichter 21 3.2 Flying Capacitor Stromrichter (FLC) 23 3.2.1 3-Level Flying Capacitor Stromrichter (3L-FLC) 23 3.2.2 4-Level Flying Capacitor-Stromrichter (4L-FLC) 33 3.2.3 Mehrstufige Flying Capacitor-Stromrichter (NL-FLC) 39 3.3 Stacked Multicell Stromrichter (SMC) 43 3.3.1 5L-Stacked Multicell Stromrichter (5L-SMC) 43 3.3.2 N-Level Stacked Multicell Umrichter (NL-SMC) 51 4 Modellierung und Auslegung der Stromrichter 59 4.1 Verlustmodell 59 4.1.1 Sperrschichttemperaturen 64 4.2 Auslegung der Leistungshalbleiter 65 4.2.1 Stromauslegung 67 4.2.2 Worst-Case Arbeitspunkte 69 4.3 Auslegung der Zwischenkreiskondensatoren 75 4.3.1 Spannungszwischenkreis 76 4.3.2 Lastseitige Strombelastung und resultierende Spannungswelligkeit im Spannungszwischenkreis 77 4.3.3 Abhängigkeit der Strombelastung und der Spannungswelligkeit im Spannungszwischenkreis vom Frequenzverhältnis mf 95 4.3.4 Netzseitige Zwischenkreiseinspeisung 97 4.3.4.1 Zwischenkreiseinspeisung mit idealisiertem Transformatormodell 98 4.3.4.2 Zwischenkreiseinspeisung mit erweitertem Transformatormodell 101 4.3.5 Simulation des Gesamtsystems 104 4.4 Auslegung der Flying Capacitors 107 4.4.1 Strombelastung der Flying Capacitors 109 4.4.2 Spannungswelligkeit über den Flying Capacitors 113 4.4.3 Abhängigkeit der Spannungswelligkeit der Flying Capacitors vom Frequenzverhältnis mf 124 4.4.4 Auswirkung der Spannungswelligkeit der Flying Capacitors auf die Ausgangsspannungen 126 5 Vergleich der Stromrichtertopologien 129 5.1 Daten für den Stromrichtervergleich 129 5.2 Basis des Vergleiches 132 5.3 Vergleich für einen 2,3 kV Mittelspannungsstromrichter 134 5.3.1 Vergleich bei verschiedenen Schaltfrequenzen 134 5.3.2 Vergleich bei maximaler Trägerfrequenz 142 5.4 Vergleich für einen 4,16 kV Mittelspannungsstromrichter 146 5.4.1 Vergleich bei verschiedenen Schaltfrequenzen 146 5.4.2 Vergleich bei maximaler Trägerfrequenz 153 5.5 Vergleich für einen 6,6 kV Mittelspannungsstromrichter 156 5.5.1 Vergleich bei verschiedenen Schaltfrequenzen 156 5.5.2 Vergleich bei maximaler Trägerfrequenz 162 5.6 Vergleich von 2,3 kV, 4,16 kV und 6,6 kV Mittelspannungsstromrichtern 165 5.6.1 Vergleich bei identischer installierter Schalterleistung SS 165 5.6.2 Vergleich bei einer identischen Ausgangsleistung 167 6 Zusammenfassung und Bewertung 171 Anhang 175 A. Halbleiterverlustmodell 175 Referenzen 177 / The thesis deals with a detailed comparison of voltage source converter topologies with a central dc-link energy storage device for medium voltage applications. The Three-Level Neutral Point Clamped Voltage Source Converter (3L-NPC VSC) is compared with multilevel Flying Capacitor (FLC) and Stacked Multicell (SMC) Voltage Source Converters (VSC) for output voltages of 2.3 kV, 4.16 kV and 6.6 kV by using state-of-the-art 6.5 kV, 3.3 kV, 4.5 kV and 1.7kV IGBTs. The fundamental functionality of the investigated converter topologies as well as the design of the power semiconductors and of the energy storage devices (Flying Capacitors and Dc-Link capacitors) is described. The installed switch power, converter losses, the semiconductor loss distribution, modulation strategies and the harmonic spectra are compared in detail.:Inhaltsverzeichnis Liste der Variablen i Liste der Abkürzungen v 1 Einleitung 1 2 Überblick von Mittelspannungsstromrichtertopologien und Leistungshalbleitern 3 2.1 Mittelspannungsumrichtertopologien 3 2.2 Leistungshalbleiter 8 3 Aufbau und Funktion von Mittelspannungsstromrichtertopologien 10 3.1 Neutral Point Clamped Stromrichter (NPC) 10 3.1.1 3-Level Neutral Point Clamped Stromrichter (3L-NPC) 10 3.1.2 Mehrstufige NPC-Umrichter 21 3.2 Flying Capacitor Stromrichter (FLC) 23 3.2.1 3-Level Flying Capacitor Stromrichter (3L-FLC) 23 3.2.2 4-Level Flying Capacitor-Stromrichter (4L-FLC) 33 3.2.3 Mehrstufige Flying Capacitor-Stromrichter (NL-FLC) 39 3.3 Stacked Multicell Stromrichter (SMC) 43 3.3.1 5L-Stacked Multicell Stromrichter (5L-SMC) 43 3.3.2 N-Level Stacked Multicell Umrichter (NL-SMC) 51 4 Modellierung und Auslegung der Stromrichter 59 4.1 Verlustmodell 59 4.1.1 Sperrschichttemperaturen 64 4.2 Auslegung der Leistungshalbleiter 65 4.2.1 Stromauslegung 67 4.2.2 Worst-Case Arbeitspunkte 69 4.3 Auslegung der Zwischenkreiskondensatoren 75 4.3.1 Spannungszwischenkreis 76 4.3.2 Lastseitige Strombelastung und resultierende Spannungswelligkeit im Spannungszwischenkreis 77 4.3.3 Abhängigkeit der Strombelastung und der Spannungswelligkeit im Spannungszwischenkreis vom Frequenzverhältnis mf 95 4.3.4 Netzseitige Zwischenkreiseinspeisung 97 4.3.4.1 Zwischenkreiseinspeisung mit idealisiertem Transformatormodell 98 4.3.4.2 Zwischenkreiseinspeisung mit erweitertem Transformatormodell 101 4.3.5 Simulation des Gesamtsystems 104 4.4 Auslegung der Flying Capacitors 107 4.4.1 Strombelastung der Flying Capacitors 109 4.4.2 Spannungswelligkeit über den Flying Capacitors 113 4.4.3 Abhängigkeit der Spannungswelligkeit der Flying Capacitors vom Frequenzverhältnis mf 124 4.4.4 Auswirkung der Spannungswelligkeit der Flying Capacitors auf die Ausgangsspannungen 126 5 Vergleich der Stromrichtertopologien 129 5.1 Daten für den Stromrichtervergleich 129 5.2 Basis des Vergleiches 132 5.3 Vergleich für einen 2,3 kV Mittelspannungsstromrichter 134 5.3.1 Vergleich bei verschiedenen Schaltfrequenzen 134 5.3.2 Vergleich bei maximaler Trägerfrequenz 142 5.4 Vergleich für einen 4,16 kV Mittelspannungsstromrichter 146 5.4.1 Vergleich bei verschiedenen Schaltfrequenzen 146 5.4.2 Vergleich bei maximaler Trägerfrequenz 153 5.5 Vergleich für einen 6,6 kV Mittelspannungsstromrichter 156 5.5.1 Vergleich bei verschiedenen Schaltfrequenzen 156 5.5.2 Vergleich bei maximaler Trägerfrequenz 162 5.6 Vergleich von 2,3 kV, 4,16 kV und 6,6 kV Mittelspannungsstromrichtern 165 5.6.1 Vergleich bei identischer installierter Schalterleistung SS 165 5.6.2 Vergleich bei einer identischen Ausgangsleistung 167 6 Zusammenfassung und Bewertung 171 Anhang 175 A. Halbleiterverlustmodell 175 Referenzen 177

The stacked polyphase bridges (SPB) converter topology is investigated in the presentthesis as a fault-tolerant choice for permanent-magnet synchronous reluctance (PMSyn-Rel) motor control. Integrated motor drive systems are studied as they offer great benefitsfor propulsion applications. Moreover, the importance of a modular topology, like theSPB, for an electric powertrain is discussed. The latter consists of a number of seriesconnected, 3-phase 2-level inverter submodules that supply separate sets of windings ina multi-star motor. The specifications of building a four-board SPB setup are examined,while the challenges of an active voltage balancing controller are analyzed. The designprocess is explained step-by-step and the final printed circuit boards (PCBs) are presented.Furthermore, the significance of low electromagnetic interference design for a converterthat requires high speed communication is highlighted. Finally, the prototype is testedthoroughly and the expected fault-tolerant capabilities are validated on a PMSynRel motor. / I detta examensarbete unders¨ok SPB-omriktartopologin (stacked polyphase bridges converter)i termer av ett feltolerant elektriskt drivsystem f¨or en permamentmagnetassisteradsynkron reluktansmaskin (PMSynRel). SPB-omriktaren best°ar av ett antal seriekoppladetrefasomriktare av tv°aniv°atyp som, var och en, f¨orser effekt till en trefaslindningtillh¨orande en modul¨ar elmaskin av multifastyp. Specifikation, design och konstruktion aven SPB-omriktare med fyra seriekopplade moduler studeras. Designproceduren presenterasi en steg-f¨or-steg-process och de tillverkade kretskorten presenteras utf¨orligt. Kommunikationenmellan de olika kretskorten unders¨oks med s¨arskild tonvikt p°a l°ag elektromagnetiskinterferens vilket ¨ar n¨odv¨andigt om h¨og kommunikationshastighet skall kunnauppn°as. Den f¨ardigst¨allda prototypen har utv¨arderats experimentellt och kapaciteten f¨orfeltolerans har demonstrerats vid drift av en PMSynRel-maskin utrustad med en multifaslindning.

Lithium ion batteries (LIBs) have attracted considerable interest due to their wide range of applications, such as portable electronics, electric vehicles (EVs) and aerospace applications. Particularly, the emergence of a variety of nanostructured materials has driven the development of LIBs towards the next generation, which is featured with high specific energy and large power density. Herein, rolled-up nanotechnology is introduced for the design of strain-released materials as anodes of LIBs. Upon this approach, self-rolled nanostructures can be elegantly combined with different functional materials and form a tubular shape by relaxing the intrinsic strain, thus allowing for enhanced tolerance towards stress cracking. In addition, the hollow tube center efficiently facilitates electrolyte mass flow and accommodates volume variation during cycling. In this context, such structures are promising candidates for electrode materials of LIBs to potentially address their intrinsic issues. This work focuses on the development of superior structures of Si and SnO2 for LIBs based on the rolled-up nanotech. Specifically, Si is the most promising substitute for graphite anodes due to its abundance and high theoretical gravimetric capacity. Combined with the C material, a Si/C self-wound nanomembrane structure is firstly realized. Benefiting from a strain-released tubular shape, the bilayer self-rolled structures exhibit an enhanced electrochemical behavior over commercial Si microparticles. Remarkably, this behavior is further improved by introducing a double-sided carbon coating to form a C/Si/C self-rolled structure. With SnO2 as active material, an intriguing sandwich-stacked structure is studied. Furthermore, this novel structure, with a minimized strain energy due to strain release, exposes more active sites for the electrochemical reactions, and also provides additional channels for fast ion diffusion and electron transport. The electrochemical characterization and morphology evolution reveal the excellent cycling performance and stability of such structures.

This project is concerned with both the study of different Multiple Quantum Wells (MQWs) structures using the In0.53Ga0.47As/In0.52Al0.48As material system lattice matched to InP and a systematic investigation of the properties of InAs QD systems within GaAs with the aim of achieving enhancements of solar cell performance. The key challenge is the growth of QDs solar cell structures which exhibit sufficient absorption (enhanced infrared absorption) to increase short circuit current density (Jsc) but which can still maintains a high open circuit voltage (Voc). The research consists of epitaxial growth using state-of–the-art MBE, optical absorption, photoluminescence and high resolution x-ray diffraction measurements as well as device fabrication and characterization of novel solar cell structures. Optimization was performed on these novel cells to further improve their efficiency by inserting stacks of QD into different regions of the device. The effect of localized doping of such structures was used in an attempt to maintain and enhance the open-circuit voltage which in turn increases the device efficiency. The fabricated devices were characterized using measurements of the dark/light current-voltage (I-V) characteristics and spectral response (50-480 K). Solar cell external quantum efficiencies under standard air mass (AM) 1.5 spectrum were determined and the suitability of these new cells under solar concentration were assessed. Full physical simulations are performed using SILVACO semiconductors modelling software to generate models of multi-junction solar cells that were crucial in informing iterations to growth and fabrication and help to reconcile theory with experiment. One of the key findings, of this thesis, is the fact that Intermediate band photovoltaic devices using material based on InAs/GaAs vertically stacked quantum dot arrays, can be used in applications according to specific configuration criteria such as high temperature operation conditions. The intermediate band cell, including an inter-dot doped configuration, has been found to be a potential candidate as the inter dot doping profile reduces the efficiency degradation below the GaAs values including an enhancement in the open circuit voltage. It has been proved that these devices not only have a good performance at high temperatures but also by changing the vertical stacking QD layer periodicity can enhance the short circuit current density while keeping a large open circuit voltage. It was confirmed in practical device operation that thermal energy is required to enable the intermediate band in InAs/GaAs QD materials. The impact of this works can help in the future improvements of the intermediate band solar cells based on InAs on GaAs QD. The best overall efficiency of 11.6 % obtained in this work is an excellent value for so simple devices configuration. The Si3N4, tested for the first time on InAs/GaAs QD materials, reduces the reflectance on the device surface to a value of 2% and the operational wavelength can be tuned by controlling the layer thickness. A 100 nm Si3N4 antireflective coating proved to be an excellent coating from 700 to 1000 nm. In terms of short circuit current density a 37% enhancement was achieved.

Organic conjugated systems are attractive because of wide range of applications, which includes stimulated emission from -conjugated polymers, optical switches, organic solar cells and organic light emitting diodes, to name a few. They have the advantage of low cost, ease of processing and tunability of their linear and nonlinear optical properties by functionalization with donor or acceptor groups. In chapter 1, we provide an introduction to the π-conjugated systems and various interesting phenomena observed in these systems, This is followed by a brief description of the application of the above mentioned -conjugated systems for organic light emitting diodes, and organic photovoltaic cells. In the last section of this chapter, we give an introduction to magnetism due to π-electron systems. In chapter 2, we begin with energy band theory in one-dimension and its drawbacks. We introduce various model Hamiltonians which incorporate electron-electron inter-actions like Hubbard model, and Pariser-Parr-Pople (PPP) model. We present numerical techniques like valence-bond (VB) and constant MS techniques that are used to exactly solve the above model Hamiltonian. This is followed by an introduction to density matrix renormalization group method (DMRG) employed for the above model Hamiltonian for larger system in one-dimension and quasi-one-dimension. We give description of linear and nonlinear optical properties followed by the oriented gas model for ensemble of molecules. Various methods for computing polarizabilities and hyperpolarizibilities of molecules includes such as Finite-Field method, Sum-Over-State Method, and Correction Vector (CV) Methods are described in detail. In chapter 3, we look into fused azulene systems as a possible organic multiferroics. Azulene molecule with fused five and seven membered π-conjugated rings has a dipole moment, and the π-framework has geometric frustration. Hence in fused azulenes we can expect both ferroelectric and magnetic ground state. To explore this, we study low-lying correlated electronic states of fused azulenes using the long-range interacting PPP model and the finite DMRG method. The ground state is a singlet for oligomers up to 5 azulene units. For oligomers with more than 5 azulene units and up to 11 azulene units the oligomers have a triplet ground state. From the excitation gaps between the lowest MS = 0 state and the lowest states in MS=1, 2, and 3 sectors we predict that the ground-state spin of the fused azulene increases with the number of azulene units. In the thermodynamic limit, we expect the fused azulene to be a ferromagnet. Charge density calculations show that the ground state of the system has ferroelectric alignment of the dipoles of the monomeric units. Thus, a fused azulene system could be the first example of an organic molecule which is both ferromagnetic and ferroelectric, in the ground state. In chapter 4, we study the linear and nonlinear optical properties of diradical systems. We have studied linear and non-linear optical properties of π-conjugated diradicals because they are expected to exhibit large non-linear responses. The system studied are oligomers of dicyclopenta-fused acenes (DPA) and the s-indaceno[1,2,3-cd;5,6,7-c'd']diphenalene (IDPL) molecule. Spin-spin correlation functions within a correlated PPP model Hamiltonian, using exact diagonalization method, are used to characterize the diradical nature of DPA-2 and similar calculations on Anthracene have been performed to contrast this with a singlet character. The diradical character of DPA-2 is also manifest as low optical gap, low spin gap and large THG coefficients compared to Anthracene molecule. Larger DPA-k, k > 2, oligomers as well as the IDPL molecule have been studied within the DMRG technique. In the DPA-4, we nd a very small spin gap (0.04 eV), while in the oligomers with k > 4, we nd that the ground state is degenerate with the lowest triplet state. The energy of the second excited triplet state decreases with increasing size k and seems to saturate at ~0.36 eV in the thermodynamic limit. The lowest optical gap in DPA-4 is at 1.94 eV and has large transition dipoles, while for DPA-k, 4 < k ≥ 28, we have not been able to access states with large transition dipoles. The weak low-energy excitations seem to saturate at 0.5 eV and the two-photon gap also seems to be saturating at~ 0.3 eV in the thermodynamic limit. These polymers will not be IR uorescent by Kasha rule. The dominant component of the THG coefficient, γxxxx, is highest for DPA-4 which reduces almost by an order of magnitude in DPA-8; for k > 8 it increases up to the largest system with k=20 for which we have computed the coe cient. The variation of the charge gap of DPA oligomer with the increase in system size is small and in the polymer limit the charge gap is 4:24 eV. For IDPL molecule spin gap is 0.20 eV and next excited triplet state is at 1.48 eV. Two lowest singlet states in B space are nearly degenerate and have large transition dipole moments. Optical gaps to the above states are 2.20 eV and 2.22 eV. Two-photon gap in this system is 1.29 eV, hence this system is also non- uorescent. We calculated the dispersion of the major component of the THG coefficient, γxxxx, over a wide frequency range for this molecule, and we observe resonances corresponding to the 21Ag and 11Bu states. Extrapolated value of γxxxx at zero frequency is 15:58 x 106 a.u which is very large and the system does not have any donor or acceptor substituent groups. In chapter 5, we study absorption spectra and two photon absorption coefficient of expanded porphyrins (EPs). We nd that in the 4n+2 EPs there are two prominent low-lying one-photon excitations while in 4n systems there is only one such excitation. The two-photon gaps in both these types of systems are at energies close to the one-photon excitations. The spin gap in 4n+2 EPs are very small although the spin-1/2 Heisenberg calculations show that a pure spin system in this geometry will not have vanishing spin-gap. The charge density rearrangement in the one-photon excited state is most at the aza nitrogen site and at the meso carbon sites. In the two-photon states also the charge density rearrangement occurs mostly at the aza-ring sites. The bond order changes in these states is much more striking. In the one-photon state, the C-C bond length in the aza rings show a tendency to become uniform. Similar qualitative trend is also observed for the two-photon state. In chapter 6, we study linear and nonlinear optical properties of two push-pull polyenes stacked in head to head (HtH) and head to tail con gurations (HtT), at different stacking angles, exactly within the PPP model. Varying the stacking angle between polyenes, we nd that the optical gap varies slightly, but transition dipoles show large variation. The dominant component of first-order hyperpolarizability, βxxx for HtH and βyyy for HtT arrangement strongly depend on the distance between molecules. The βxxx for HtH configuration shows a maximum at a nonzero stacking angle, which varies with inter polyene distance. ZINDO study on two monomers, (4-hydroxy-40-nitro-azobenzene) connected by a conjugated bridge shows that βav is more than twice the monomer value and with a red-shift in the optical gap. In chapter 7, we have calculated the shifts in optical gaps and band edges as a function of the distance between two monomers within a correlated PPP model Hamiltonian for various stacking geometries. We have used as model monomers, both unsubstituted polyenes and push-pull substituted polyenes. We have carried out calculations with and without inter-chain hopping between sites on different molecules. We note that in the absence of inter-chain hopping, the energy level shifts are almost independent of the distances between the chains in all stacking geometries. It is also interesting to note that only electron-electron interactions yield a blue shift in the optical gaps for parallel stacking, but red shift in the gap for all other stacking geometries. We note that most of the shift in the gap is due to shifts in the excited state energy and the ground-state energy remains almost the same. With interchain transfer the shift in the optical gap increases with decrease in the interchain distance. We observe red-shifts in parallel stacking geometry when inter-chain electron hopping is turned on, at small interchain separations. In general interchain hopping increases significantly the red shift in the optical gaps for all geometries. Even for push-pull polyenes of | e| =2.0 eV, we observe the same trend in the shift in the optical gap for various stacking geometries. In this case the shift in optical gap is an order of magnitude higher when interchain hopping is turned on compared to that in the absence of interchain hopping. We find that the optical gap shifts are largest for the parallel stacking geometry, and it also shows stronger distance dependence. This is in close conformity with experimental observation of red-shift in absorption maxima when hydrostatic pressure is applied on the system. The shift in the HOMO (LUMO) level is small in the absence of t?, and the largest shift is in the case of parallel stacking compared to other stacking geometries. The distance dependence of the HOMO shifts is also rather weak. When t? is turned on, the level shifts become large by a factor of five or more. When we have push-pull groups electron-hole symmetry is broken and the shift is different for the HOMO and the LUMO level. Depending upon stacking geometry, the HOMO shifts vary from 0:1 ~ eV to 0.3 eV, which is larger than the shifts observed in unsubstituted polyenes. This large shift in the LUMO reduces the efficiency of exciton dissociation.

This diploma thesis deals with the design and simulation of multi-layer stacked Yagi antenna at the frequency of 60 GHz. At work are proposed and compared two ways of antenna feeding methods. Based on the results obtained, antenna is fed by substrate integrated waveguide. The proposed antenna has been optimized in terms of gain and impedance matching at the specified frequency. The correct functionality of the antenna was verified by simulation program CST MWS 2011. Simulation results were verified experimentally.

Digital twins for industrial control systems have gained significant interest over recent years. This attention is mainly because of the advanced capabilities offered by digital twins in the areas of simulation, optimization, and predictive maintenance. Some recent studies discuss the possibility of using digital twins for intrusion detection in industrial control systems. To this end, this thesis aims to propose a security framework for industrial control systems including its digital twin for security monitoring and a machine learning-based intrusion detection system for real-time intrusion detection. The digital twin solution used in this study is a standalone simulation of an industrial filling plant available as open-source. After thoroughly evaluating the implementation aspects of the existing knowledge-driven open-source digital twin solutions of industrial control systems, this solution is chosen. The cybersecurity analysis approach utilizes this digital twin to model and execute different realistic process-aware attack scenarios and generate a training dataset reflecting the process measurements under normal operations and attack scenarios. A total of 23 attack scenarios are modelled and executed in the digital twin and these scenarios belong to four different attack types, naming command injection, network DoS, calculated measurement injection, and naive measurement injection. Furthermore, the proposed framework also includes a machine learning-based intrusion detection system. This intrusion detection system is designed in two stages. The first stage involves an offline evaluation of the performance of eight different supervised machine learning algorithms on the labelled dataset. In the second stage, a stacked ensemble classifier model that combines the best performing supervised algorithms on different training dataset labels is modelled as the final machine learning model. This stacked ensemble model is trained offline using the labelled dataset and then used for classifying the incoming data samples from the digital twin during the live operation of the system. The results show that the designed intrusion detection system is capable of detecting and classifying intrusions in near real-time (0.1 seconds). The practicality and benefits of the proposed digital twin-based security framework are also discussed in this work. / Digitala tvillingar för industriella styrsystem har fått ett betydande intresse under de senaste åren. Denna uppmärksamhet beror främst på de avancerade möjligheter som digitala tvillingar erbjuder inom simulering, optimering och förutsägbart underhåll. Några färska studier diskuterar möjligheten att använda digitala tvillingar för intrångsdetektering i industriella styrsystem. För detta ändamål syftar denna avhandling till att föreslå ett säkerhetsramverk för industriella styrsystem inklusive dess digitala tvilling för säkerhetsövervakning och ett maskininlärningsbaserat intrångsdetekteringssystem för intrångsdetektering i realtid. Den digitala tvillinglösningen som används i denna studie är en fristående simulering av en industriell fyllningsanläggning som finns tillgänglig som öppen källkod. Efter noggrann utvärdering av implementeringsaspekterna för de befintliga kunskapsdrivna digitala tvillinglösningarna med öppen källkod för industriella styrsystem, väljs denna lösning. Cybersäkerhetsanalysmetoden använder denna digitala tvilling för att modellera och exekvera olika realistiska processmedvetna attackscenarier och generera en utbildningsdataset som återspeglar processmätningarna under normala operationer och attackscenarier. Totalt 23 angreppsscenarier modelleras och utförs i den digitala tvillingen och dessa scenarier tillhör fyra olika angreppstyper, namnskommandoinjektion, nätverks -DoS, beräknad mätinjektion och naiv mätinjektion. Dessutom innehåller det föreslagna ramverket också ett maskininlärningsbaserat system för intrångsdetektering. Detta intrångsdetekteringssystem är utformat i två steg. Det första steget innebär en offline -utvärdering av prestanda för åtta olika algoritmer för maskininlärning övervakad på den märkta datauppsättningen. I det andra steget modelleras en staplad ensemble -klassificerarmodell som kombinerar de bäst presterande övervakade algoritmerna på olika etiketter för utbildningsdataset som den slutliga modellen för maskininlärning. Denna staplade ensemblemodell tränas offline med hjälp av den märkta datauppsättningen och används sedan för att klassificera inkommande dataprover från den digitala tvillingen under systemets levande drift. Resultaten visar att det konstruerade intrångsdetekteringssystemet kan upptäcka och klassificera intrång i nära realtid (0,1 sekunder). Det praktiska och fördelarna med den föreslagna digitala tvillingbaserade säkerhetsramen diskuteras också i detta arbete.

We consider a mechanistic non-linear machine learning approach to learning signals in financial time series data. A modularised and decoupled algorithm framework is established and is proven on daily sampled closing time-series data for JSE equity markets. The input patterns are based on input data vectors of data windows preprocessed into a sequence of daily, weekly and monthly or quarterly sampled feature measurement changes (log feature fluctuations). The data processing is split into a batch processed step where features are learnt using a Stacked AutoEncoder (SAE) via unsupervised learning, and then both batch and online supervised learning are carried out on Feedforward Neural Networks (FNNs) using these features. The FNN output is a point prediction of measured time-series feature fluctuations (log differenced data) in the future (ex-post). Weight initializations for these networks are implemented with restricted Boltzmann machine pretraining, and variance based initializations. The validity of the FNN backtest results are shown under a rigorous assessment of backtest overfitting using both Combinatorially Symmetrical Cross Validation and Probabilistic and Deflated Sharpe Ratios. Results are further used to develop a view on the phenomenology of financial markets and the value of complex historical data under unstable dynamics.

Using automatic methods to assign incoming tracks and albums from multiple sources to artists entities in a digital rights management company, where no universal artist identifier is available and artist names can be ambiguous, is a challenging problem. In this work we propose to use stacked generalization to combine the predictions of heterogeneous classifiers for an improved quality of artist content matching on two datasets from a digital rights management company. We compare the performance of using a nonlinear meta-level learner to a linear meta-level learner for the stacked generalization on the two datasets, as well as on eight additional datasets to see how well our results general- ize. We conduct experiments and evaluate how the different meta-level learners perform, using the base learners’ class probabilities or a combination of the base learners’ class probabilities and original input features as meta-features. Our results indicate that stacking with a non-linear meta-level learner can improve predictions on the artist chooser problem. Furthermore, our results indicate that when using a linear meta-level learner for stacked generalization, using the base learners’ class probabilities as metafeatures works best, while using a combination of the base learners’ class probabilities and the original input features as meta-features works best when using a non-linear metalevel learner. Among all the evaluated stacking approaches, stacking with a non-linear meta-level learner, using a combination of the base learners’ class probabilities and the original input features as meta-features, performs the best in our experiments over the ten evaluation datasets. / Att använda automatiska metoder för att tilldela spår och album från olika källor till artister i en digital underhållningstjänst är problematiskt då det inte finns några universellt använda identifierare för artister och namn på artister kan vara tvetydiga. I det här verket föreslår vi en användning av staplad generalisering för att kombinera förutsägningar från heterogena klassificerare för förbättra artistmatchningen i två datamäng från en digital underhållningstjänst. Vi jämför prestandan mellan en linjär och en icke-linjär metainlärningsmetod för den staplade generaliseringen av de två datamängder, samt även åtta ytterligare datamäng för att se hur resultaten kan generaliseras. Vi utför experiment och utvärderar hur de olika metainlärningsmetoderna presterar genom att använda basinlärningsmetodens klassannolikheter eller en kombination av basinlärningsmetodens klassannolikheter och den ursprungliga representationen som metarepresentation. Våra resultat indikerar att staplandet med en icke-linjär metainlärningsmetod kan förbättra förutsägningarna i problemet med att tilldela artister. Vidare indikerar våra resultat att när man använder en linjär metainlärningsmetod för en staplad generalisering är det bäst att använda basinlärningsmetodens klassannolikheter som metarepresentation, medan när man använder en icke-linjär metainlärningsmetod för en staplade generaliseringen är det bäst att använda en kombination av basinlärningsmetodens klassannolikheter och den ursprungliga representationen som metarepresentation. Av alla utvärderade sätt att stapla är staplandet med en icke-linjär metainlärningsmetod med en kombination av basinlärningsmetodens klassannolikheter och den ursprungliga representationen som metarepresentation den ansats som presterar bäst i våra experiment över de tio datamängderna.

Bei der vierstimmigen Chanson En venant de Lyon von Jean Mouton ist nur die unterste Stimme notiert; die restlichen drei Stimmen ergeben sich durch einen sukzessiven Strettakanon (einen sogenannten stacked canon) in der Oberquart im Semibrevis-Abstand. Die Dux-Fortschreitungsregeln für diesen Kanontyp (und andere Engführungskanons mit beliebiger Stimmenzahl und beliebigen Einsatzintervallen) können algorithmisch hergeleitet werden. Anhand dieser Fortschreitungsregeln kann ein anderer Algorithmus mit Hilfe eines Zufallsgenerators Kanons erzeugen. In diesem Aufsatz wird der Frage nachgegangen, ob man die für stacked canons in der Oberquart algorithmisch generierten Fortschreitungsregeln zu einem für die Improvisation praxistauglichen Regelsystem reduzieren kann. Außerdem wird untersucht, inwiefern stacked canons in der Oberquart mit mehr als vier Stimmen möglich und ästhetisch sinnvoll sind. / In the four-part chanson En venant de Lyon by Jean Mouton, only the lowest part is notated; the other three parts arise via a so-called stacked canon, with each successive voice entering a semibreve later and a fourth above. The progression rules for the dux voice in this canon type (and in other stretto canons with any number of parts and any interval between the parts) can be derived by an algorithm. Using these progression rules, another algorithm can produce randomly generated canons. For stacked canons at the upper fourth, this paper investigates whether one can streamline the algorithmically generated rules into practical rules suitable for improvisation. Furthermore it is examined whether it is possible to generate aesthetically pleasing stacked canons at the fourth with more than four voices.