With the development of electric vehicle (EV) charging systems, energy storage systems (ESS), data center power supplies, and solid-state transformer (SST) systems, the fixed-ratio isolated DC-DC converter, namely the DC transformer (DCX), has gained significant popularity. Similar to the passive AC transformer, DCX can bidirectionally convey DC power with very high efficiency. Due to zero-voltage switching (ZVS) and a small root mean square (RMS) current, the open-loop CLLC resonant converter operating at the resonant frequency is a promising candidate for DCX with a constant voltage transfer ratio. In Chapter 2, to solve unsmooth bidirectional power flow and current distortion in the traditional CLLC-DCX with synchronization rectification (SR) modulation, a dual-active-synchronization (DAS) modulation is adopted with identical driving signals on both sides. First, the switching transition of this modulation is thoroughly analyzed considering the large switch's output capacitances. After comparing different transitions, a so-called sync-ZVS transition is more desirable with ZVS, has no deadtime conduction loss, and almost has load-independent voltage gain. An axis and center symmetric (ACS) method is proposed to achieve this switching transition. Based on this method, an overall design procedure of CLLC-DCX with DAS modulation is also proposed. However, designing a high-power and high-frequency transformer for CLLC-DCX presents significant challenges due to the trade-off between thermal management, leakage inductance minimization, and insulation requirements. To overcome this trade-off between power rating and operation frequency, a scalable electronic-embedded transformer (EET) with a low-voltage bridge integrated into the transformer windings is proposed in Chapter 3. The EET addresses the challenge through simple open-loop control and natural current sharing, enabling easy parallel connection and scaling to different power ratings. Based on this concept, a bidirectional, EET-based DC transformer (EET-DCX) is proposed to solve the transformer-level paralleling and resonant point shift issues in traditional LLC-DCX designs. By employing the embedded full bridge, the EET-DCX effectively cancels out the impedance of the leakage inductance, ensuring optimal operation at any frequency. Additionally, the EET-DCX retains the inherent advantages of the LLC-DCX, such as load-independent voltage gain, simple open-loop control, full-load range ZVS, and low circulating current. Leveraging these advantages, the proposed EET-DCX solution has the potential to push the boundaries of transformer performance to the MHz operation frequency range with hundreds of kilowatts of power capability. Moreover, to address the significant RMS current problem of the CLLC-DCX, a trapezoidal current modulation is also proposed in Chapter 3. Compared to the CLLC-DCX with a sinusoidal current, an EET-DCX with a trapezoidal current can reduce the total conduction loss by up to 23%. This total conduction loss includes semiconductor loss on both high-voltage and low-voltage bridges and transformer winding loss. In light of this EET concept, another resonant commutation (RC) EET-DCX is proposed to streamline the circuit. First, it replaces the embedded full bridge with a low-voltage bidirectional AC switch. Second, it introduces a resonant current commutation to realize a quasi-trapezoidal transformer current with a smaller RMS value. Compared to the triangular current produced by the original EET-DCX, the RMS current can be decreased by 15%. By incorporating only one embedded bidirectional AC switch, the high-frequency transformer leakage inductance impedance is fully neutralized. As a result, the rated power of the proposed RC EET-DCX can be readily scaled up through transformer-level parallelism. Furthermore, the RC EET-DCX maintains the benefits of a typical LLC/CLLC-DCX, including load-independent voltage gain, full load range ZVS, and low circulating current. However, either in EET-DCX or RC EET-DCX, the trapezoidal current modulation will increase the voltage stress on the low-voltage full bridge or bidirectional AC switch, especially when the leakage inductance is large and variable, such as in the high-power wireless charging application. To address this trade-off between RMS current and voltage stress, this paper proposes the concept of a hybrid resonant-type EET-DCX with a series resonant capacitor. Following this concept, two specific topologies, hybrid EET-DCX and hybrid RC EET-DCX, are proposed. The main difference between these topologies is that the former adopts a full bridge. In a hybrid RC EET-DCX, a resonant current commutation scheme is developed. Among these topologies, since the passive capacitor can mainly cancel the leakage inductance impedance, the full bridge or AC switch only needs to handle the remaining impedance. Thus, the voltage stress on active components can be dramatically decreased. Additionally, these two proposed topologies can retain all the advantages of previous EET-DCX designs, including natural current sharing, load-independent voltage gain, simple open-loop control, and full-load range ZVS. The comparison between these two topologies is thoroughly studied. Finally, a 12-kW DCX testbench is built to verify all the analysis and performance in Chapter 3. If output voltage regulation is required, DCX can cooperate with other voltage regulators to realize high conversion efficiency and power density. In Chapter 4, two DCX applications are implemented: an 18-kW 98.8% peak efficiency EV battery charger with partial power processing and a 50-kW symmetric 3-level buck-boost converter with common-mode (CM) noise reduction. In the first battery charger, a large portion of the power is handled by an 18 kW CLLC-DCX, and the remaining partial power goes through a 3-phase interleaved buck converter. The proposed switching transition optimization in Chapter 2 is adopted in the 18-kW CLLC-DCX to realize 98.8% peak efficiency. To handle the step-up and step-down cases at the same time, a symmetric 3-level buck-boost converter with coupled inductors is also studied as a post regulator. With symmetric topology and quadrangle current control, the converter can achieve a CM noise reduction and full load range ZVS with a small RMS current. To further optimize the performance and simplify the control, a mid-point bridging with a better CM noise reduction and a split capacitor voltage auto-balance is implemented. A 50-kW prototype is built to verify the above analysis. To summarize, Chapter 2 first proposes a switching transition optimization for CLLC-DCX. Later, to address the intrinsic trade-off between transformer rating power and frequency, an EET concept and its corresponding soft-switching DCX family are found in Chapter 3. Finally, to handle voltage regulation, two examples for practical applications are studied in Chapter 4 —one is an 18-kW partial power converter, and the other is a 50-kW 3-L buck-boost converter. Finally, Chapter 5 will draw conclusions and illustrate future work. / Doctor of Philosophy / With the development of electric vehicle (EV) charging systems, energy storage systems (ESS), data center power supply, and solid-state transformer (SST) systems, the fixed-ratio isolated dc-dc converter, namely dc transformer (DCX), has gained significant popularity. However, designing a high-performance DCX still has many challenges, such as large dead time loss, poor current sharing, and sensitivity to parameter tolerance. Firstly, the state-of-the-art resonant CLLC-DCX is optimized in Chapter 2. With an optimal switching frequency and dead time, both the primary and secondary sides of zero voltage switching (ZVS) can begin and finish simultaneously, which means dead time loss caused by current through the body diode can be eliminated. Therefore, the efficiency of CLLC-DCX can be improved. However, designing a high-power and high-frequency CLLC-DCX transformer still presents significant challenges due to the trade-off between thermal management, leakage inductance minimization, and insulation requirements. To overcome this trade-off, in Chapter 3, a scalable electronic-embedded transformer (EET) concept with a low-voltage bridge integrated into the transformer windings is proposed. The EET addresses the challenge through its simple open-loop control and natural current sharing, enabling easy parallel connection and scaling to different power ratings. In light of this EET concept, a new family of soft-switching DCXs is proposed for different applications, such as high-power wireless charging systems. All these EET-based DCXs retain the merits of typical CLLC-DCX, such as small circulating current ringing, small turn-off current, full load range ZVS, and load-independent gain. After realizing a desirable design for DCX, Chapter 4 presents two DCX applications with voltage regulation. Firstly, an 18 kW 98.8% peak efficiency battery charger is designed with partial power processing. Most of the power will go through an optimized DCX, and the remaining small portion of power will go through a 3-phase interleaved buck converter. On the other hand, DCX can also be adopted as a front-end or rear-end converter in a typical two-state DC-DC converter. As for another stage, a non-isolated DC-DC converter with a large output range can be used to handle voltage regulation. Following this structure, a 50-kW symmetric 3-L buck-boost converter with coupled inductors and reduced common emission is proposed. To summarize, the state-of-the-art CLLC-DCX is optimized in Chapter 2. Afterward, a new concept of EET-DCX and its corresponding DCX family is proposed in Chapter 3. After obtaining an optimized DCX, two practical applications with DCX are implemented in Chapter 4. Finally, Chapter 5 will draw conclusions and illustrate future work.

As modern systems are becoming increasingly complex, they are also becoming more and more cumbersome to diagnose and fix when things go wrong. One domain where it is very important for machinery and equipment to stay functional is in the world of medical IT, where technology is used to improve healthcare for people all over the world. This thesis aims to help with reducing downtime on critical life-saving equipment by implementing automatic analysis of system logs that without any domain experts involved can give an indication of the state that the system is in. First, a literature study was performed where three potential candidates of suitable neural network architectures was found. Next, the networks were implemented and a data pipeline for collecting and labeling training data was set up. After training the networks and testing them on a separate data set, the best performing model out of the three was based on GRU (Gated Recurrent Unit). Lastly, this model was tested on some real world system logs from two different sites, one without known issues and one with slow image import due to network issues. The results showed that it was feasible to build such a system that can give indications on external parameters such as network speed, latency and packet loss percentage using only raw system logs as input data. GRU, 1D-CNN (1-Dimensional Convolutional Neural Network) and Transformer's Encoder are the three models that were tested, and the best performing model was shown to produce correct patterns even on the real world system logs. / I takt med att moderna system ökar i komplexitet så blir de även svårare att felsöka och reparera när det uppstår problem. Ett område där det är mycket viktigt att maskiner och utrustning fungerar korrekt är inom medicinsk IT, där teknik används för att förbättra hälso- och sjukvården för människor över hela världen. Syftet med denna avhandling är att bidra till att minska tiden som kritisk livräddande utrustning inte fungerar genom att implementera automatisk analys av systemloggarna som utan hjälp av experter inom området kan ge en indikation på vilket tillstånd som systemet befinner sig i. Först genomfördes en litteraturstudie där tre lovande typer av neurala nätverk valdes ut. Sedan implementerades dessa nätverk och det sattes upp en datapipeline för insamling och märkning av träningsdata. Efter att ha tränat nätverken och testat dem på en separat datamängd så visade det sig att den bäst presterande modellen av de tre var baserad på GRU (Gated Recurrent Unit). Slutligen testades denna modell på riktiga systemloggar från två olika sjukhus, ett utan kända problem och ett där bilder importerades långsamt på grund av nätverksproblem. Resultaten visade på att det är möjligt att konstruera ett system som kan ge indikationer på externa parametrar såsom nätverkshastighet, latens och paketförlust i procent genom att enbart använda systemloggar som indata.  De tre modeller som testades var GRU, 1D-CNN (1-Dimensional Convolutional Neural Network) och Transformer's Encoder. Den bäst presterande modellen visade sig kunna producera korrekta mönster även för loggdata från verkliga system.

During an investigation carried out by a police officer or a gendarme, audition reports are written, the length of which can be up to several pages. The high-level goal of this thesis is to study various automatic and reliable text summarization methods to help with this time-consuming task. One challenge comes from the specific, French and judicial data that we wish to summarize; and another challenge comes from the need for reliable and factual models. First, this thesis focuses on automatic summarization evaluation, in terms of both content (how well the summary captures essential information of the source text) and factuality (to what extent the summary only includes information from or coherent with the source text). Factuality evaluation, in particular, is of crucial interest when using LLMs for judicial purposes, because of their hallucination risks. Notably, we propose a light variation of SelfCheckGPT, which has a stronger correlation with human judgment (0.743) than the wide-spread BARTScore (0.542), or our study dataset. Other paradigms, such as Question-Answering, are studied in this thesis, which however underperform compared to these. Then, extractive summarization methods are explored and compared, including one based on graphs via the TextRank algorithm, and one based on greedy optimization. The latter (overlap rate: 0.190, semantic similarity: 0.513) clearly outperforms the base TextRank (overlap rate: 0.172, semantic similarity: 0.506). An improvement of the TextRank with a threshold mechanism is also proposed, leading to a non-negligible improvement (overlap rate: 0.180, semantic similarity: 0.513). Finally, abstractive summarization, with pre-trained LLMs based on a Transformer architecture, is studied. In particular, several general-purpose and multilingual models (Llama-2, Mistral and Mixtral) were objectively compared on a summarization dataset of judicial procedures from the French police. Results show that the performances of these models are highly related to their size: Llama-2 7B struggles to adapt to uncommon data (overlap rate: 0.083, BARTScore: -3.099), while Llama-2 13B (overlap rate: 0.159, BARTScore: -2.718) and Llama-2 70B (overlap rate: 0.191, BARTScore: -2.479) have proven quite versatile and efficient. To improve the performances of the smallest models, empirical prompt-engineering and parameter-efficient fine-tuning are explored. Notably, our fine-tuned version of Mistral 7B reaches performances comparable to those of much larger models (overlap rate: 0.185, BARTScore: -2.060), without the need for empirical prompt-engineering, and with a linguistic style closer to what is expected. / Under en utredning som görs av en polis eller en gendarm skrivs förhörsprotokoll vars längd kan vara upp till flera sidor. Målet på hög nivå med denna rapport är att studera olika automatiska och tillförlitliga textsammanfattningsmetoder för att hjälpa till med denna tidskrävande uppgift. En utmaning kommer från de specifika franska och rättsliga uppgifter som vi vill sammanfatta; och en annan utmaning kommer från behovet av pålitliga, sakliga och uppfinningsfria modeller. För det första fokuserar denna rapport på automatisk sammanfattningsutvärdering, både vad gäller innehåll (hur väl sammanfattningen fångar väsentlig information i källtexten) och fakta (i vilken utsträckning sammanfattningen endast innehåller information från eller överensstämmer med källtexten). Faktautvärdering, i synnerhet, är av avgörande intresse när man använder LLM för rättsliga ändamål, på grund av deras hallucinationsrisker. Vi föreslår särskilt en lätt variant av SelfCheckGPT, som har en starkare korrelation med mänskligt omdöme (0,743) än den utbredda BARTScore (0,542), eller vår studiedatauppsättning. Andra paradigm, såsom Question-Answering, studeras i denna rapport, som dock underpresterar jämfört med dessa. Sedan utforskas och jämförs extraktiva sammanfattningsmetoder, inklusive en baserad på grafer via TextRank-algoritmen och en baserad på girig optimering. Den senare (överlappning: 0,190, semantisk likhet: 0,513) överträffar klart basen TextRank (överlappning: 0,172, semantisk likhet: 0,506). En förbättring av TextRank med en tröskelmekanism föreslås också, vilket leder till en icke försumbar förbättring (överlappning: 0,180, semantisk likhet: 0,513). Slutligen studeras abstrakt sammanfattning, med förutbildade LLM baserade på en transformatorarkitektur. I synnerhet jämfördes flera allmänna och flerspråkiga modeller (Llama-2, Mistral och Mixtral) objektivt på en sammanfattningsdatauppsättning av rättsliga förfaranden från den franska polisen. Resultaten visar att prestandan för dessa modeller är starkt relaterade till deras storlek: Llama-2 7B kämpar för att anpassa sig till ovanliga data (överlappning: 0,083, BARTScore: -3,099), medan Llama-2 13B (överlappning: 0,159, BARTScore: -2,718) och Llama-2 70B (överlappning: 0,191, BARTScore: -2,479) har visat sig vara ganska mångsidiga och effektiva. För att förbättra prestandan för de minsta modellerna utforskas empirisk prompt-teknik och parametereffektiv finjustering. Noterbart är att vår finjusterade version av Mistral 7B når prestanda som är jämförbara med de för mycket större modeller (överlappning: 0,185, BARTScore: -2,060), utan behov av empirisk prompt-teknik och med en språklig stil som ligger närmare vad som förväntas.

In the field of video surveillance, vast amounts of data are gathered each day. To be able to identify what occurred during a recorded session, a human annotator has to go through the footage and annotate the different events. This is a tedious and expensive process that takes up a large amount of time. With the rise of machine learning and in particular deep learning, the field of both image and video captioning has seen large improvements. Contrastive Language-Image Pretraining is capable of efficiently learning a multimodal space, thus able to merge the understanding of text and images. This enables visual features to be extracted and processed into text describing the visual content. This thesis presents a system for extracting and ranking important events from surveillance videos as well as a way of automatically generating a description of the event. By utilizing the pre-trained models X-CLIP and GPT-2 to extract visual information from the videos and process it into text, a video captioning model was created that requires very little training. Additionally, the ranking system was implemented to extract important parts in video, utilizing anomaly detection as well as polynomial regression. Captions were evaluated using the metrics BLEU, METEOR, ROUGE and CIDEr, and the model receives scores comparable to other video captioning models. Additionally, captions were evaluated by experts in the field of video surveillance, who rated them on accuracy, reaching up to 62.9%, and semantic quality, reaching 99.2%. Furthermore the ranking system was also evaluated by the experts, where they agree with the ranking system 78% of the time. / Inom videoövervakning samlas stora mängder data in varje dag. För att kunna identifiera vad som händer i en inspelad övervakningsvideo så måste en människa gå igenom och annotera de olika händelserna. Detta är en långsam och dyr process som tar upp mycket tid. Under de senaste åren har det setts en enorm ökning av användandet av olika maskininlärningsmodeller. Djupinlärningsmodeller har fått stor framgång när det kommer till att generera korrekt och trovärdig text. De har också använts för att generera beskrivningar för både bilder och video. Contrastive Language-Image Pre-training har gjort det möjligt att träna en multimodal rymd som kombinerar förståelsen av text och bild. Detta gör det möjligt att extrahera visuell information och skapa textbeskrivningar. Denna master uppsatts beskriver ett system som kan extrahera och ranka viktiga händelser i en övervakningsvideo samt ett automatiskt sätt att generera beskrivningar till dessa. Genom att använda de förtränade modellerna X-CLIP och GPT-2 för att extrahera visuell information och textgenerering, har en videobeskrivningsmodell skapats som endast behöver en liten mängd träning. Dessutom har ett rankingsystem implementerats för att extrahera de viktiga delarna i en video genom att använda anomalidetektion och polynomregression. Video beskrivningarna utvärderades med måtten BLEU, METOER, ROUGE och CIDEr, där modellerna får resultat i klass med andra videobeskrivningsmodeller. Fortsättningsvis utvärderades beskrivningarna också av experter inom videoövervakningsområdet där de fick besvara hur bra beskrivningarna var i måtten: beskrivningsprecision som uppnådde 62.9% och semantisk kvalité som uppnådde 99.2%. Ranknignssystemet utvärderades också av experterna. Deras åsikter överensstämde till 78% med rankningssystemet.

In this thesis, power system transients are considered as an opportunity for development of on-line diagnostics of power components and specifically the insulation systems of power transformers and bushings. A new technique for on-line dielectric response measurement of power transformer bushings is proposed which utilizes natural transients in the power system, such as lightning and switching surges, as stimuli. Laboratory investigations are done on implementation of the proposed technique. Measurement considerations, data acquisition and processing involved in achievement of reasonable accuracy in the Dielectric Response (DR) are presented. Capability of the technique in tracking of the degradation signatures such as moisture content in the insulation has been evaluated and it has shown a good level of accuracy by being compared to the Frequency Domain Spectroscopy (FDS).  The proposed technique is tested on the service-aged 150 kV bushings and feasibility of the technique for monitoring of dielectric properties of power transformer bushings has been assessed; the results are promising for the technique to be used in the real application.  Partial Discharges (PD) behavior under transients has been also studied for different materials in this project. PD behavior of different defects, at different insulation condition, responding to the overvoltage transients in form of superimposed impulses on ac voltages was investigated and it was perceived how their distinctive response and the interpretation of  that, can be useful for their identification. Besides the conventional materials, surface ac PD properties of modified paper with silica and zinc oxide nanoparticles under the superimposed impulses have been assessed in this project. Proper type and optimum concentration level of nanoparticles in the paper are the factors that lead to the improvement of PD behavior in the modified paper under overvoltage transients. / <p>QC 20160525</p>

Thesis (MScEng (Electrical and Electronic Engineering))--Stellenbosch University, 2011. / ENGLISH ABSTRACT: This thesis investigates the viability of constructing a solid-state transformer (SST) with a series-input, parallel-output connection of full-bridge, three-level ying-capacitor converters. It focusses on the active recti er front-end of the SST which is used to control the input current to be sinusoidal and in-phase with the sinusoidal input voltage. A stack of two converters are built and tested. The input current, as well as the ying capacitor voltages of the two active recti ers in the stack, are actively controlled by a nite-state model-based predictive (FS-MPC) controller. The use of multiple ying-capacitor converters poses a problem when using FS-MPC because of the large number of possible switching states to include in the prediction equations. Three FS-MPC control algorithms are proposed to attempt to overcome the problem associated with the large number of switching states. They are implemented on an FPGA digital controller. The algorithms are compared on the bases of voltage and current errors, as well as their responses to disturbances that are introduced into the system. The simulation and experimental results that are presented shows that by interleaving the control actions for the two converters, one can obtain fast and robust responses of the controlled variables. The viability of extending the interleaving control algorithm beyond two converters is also motivated. / AFRIKAANSE OPSOMMING: Hierdie tesis ondersoek die moontlikheid van volbrug, drievlak vlieënde-kapasitoromsetters wat gebruik word om 'n serie-intree, parallel-uittree drywingselektroniese transformator (DET) te bou. Dit fokus op die aktiewe gelykrigter van die DET wat gebruik word om die intreestroom te beheer om sinusvormig en in fase met die sinusvormige intreespanning te wees. 'n Stapel van twee omsetters word gebou en getoets. Die intreestroom, sowel as die vlieënde kapasitorspannings van die twee aktiewe gelykrigters in die stapel, word aktief beheer met behulp van 'n eindige-toestand, model-gebaseerde voorspellende beheerder (ET-MVB). Die gebruik van veelvuldige vlieënde-kapasitoromsetters bemoeilik die implementering van 'n ET-MVB-beheerder as gevolg van die groot aantal skakeltoestande wat in die voorspellende vergelykings in ag geneem moet word. Drie ET-MVB-algoritmes word voorgestel om te poog om die probleme, wat met die groot aantal skakeltoestande geassosieer word, te oorkom. Die algoritmes word in 'n FPGA digitale verwerker geïmplementeer. Die algoritmes word vergelyk op grond van hul stroom- en spanningsfoute, asook hul reaksie op steurings wat op die stelsel ingevoer word. Die simulasie en praktiese resultate toon dat, deur die beheeraksies vir die twee omsetters te laat oorvleuel, die gedrag van die beheerde veranderlikes vinniger en meer robuust is. Die moontlikheid om die oorvleuelende beheeraksies uit te brei tot meer as twee omsetters word ook gemotiveer.

In order to develop more economical and ecologically friendly transformers it is necessary to know the losses throughout the product development process. There are several losses related to transformers, but in this particular case the focus will be on the resistive loss of the transformer. In order to measure this loss first the resonant frequency of the transformer is determined. Since at resonance the secondary side of the transformer is considered to be purely resistive. The aim of this paper is to design and build a closed loop measurement system that is able to perform this task.

The objective of this Thesis is to investigate the phenomenon of voltage unbalance on electrical wind power systems. A large part of this work is the literature review of all relative work that has been done so far. This serves first as a guideline to define and measure voltage unbalance and second as a tool to spot open research questions that can inspire future work. A case study is then used to investigate the voltage unbalance at a wind farm in Näsudden, Gotland. Using real-time measurements and a simulation of the power system in MATLAB/Simulink, an evaluation of the propagation of the voltage unbalance from the distribution to the turbine level is carried out. The effect that different transformer connection types have on the propagation is studied through simulations. Many assumptions and simplifications had to be made due to several limiting factors during this work, mainly related to time and data restrictions. The main result shows that when Delta – Wye Grounded and Wye – Wye Grounded transformers are used, the unbalance is halved when it passes to the turbine side. On the other hand, when Wye Grounded – Wye Grounded configuration was used, the unbalance was unaffected. The results also include a comparison of the use of different indices to quantify a voltage unbalance.

This dissertation thesis is focused on the evaluation of the dissolved gas analysis in the power oil transformer diagnosis. This thesis derives from known, in standards, directives and literature shows realities which are fill in new pieces of knowledge as determining of gases important for evaluation of the transformer condition, specifying of current methods accuracy and investigating of accuracy change for interpretation of dissolved gases. The part of this thesis is the application of mathematic methods for detailed description of individual transformer failures and determining of dominant gas which are generating in power oil transformer. Obtain conclusion will be contribution for transformer diagnostics and will be applicable in industrial practice.

Renewable energy is harnessed from continuously replenishing natural processes. Some commonly known are sunlight, water, wind, tides, geothermal heat and various forms of biomass. The focus on renewable energy has over the past few decades intensified greatly. This thesis contributes to the research on developing renewable energy technologies, within the wind power, wave power and marine current power projects at the division of Electricity, Uppsala University. In this thesis grid connection of permanent magnet generator based renewable energy sources is evaluated. A tap transformer based grid connection system has been constructed and experimentally evaluated for a vertical axis wind turbine. Full range variable speed operation of the turbine is enabled by using the different step-up ratios of a tap transformer. This removes the need for a DC/DC step or an active rectifier on the generator side of the full frequency converter and thereby reduces system complexity. Experiments and simulations of the system for variable speed operation are done and efficiency and harmonic content are evaluated.  The work presented in the thesis has also contributed to the design, construction and evaluation of a full-scale offshore marine substation for wave power intended to grid connect a farm of wave energy converters. The function of the marine substation has been experimentally tested and the substation is ready for deployment. Results from the system verification are presented. Special focus is on the transformer losses and transformer in-rush currents. A control and grid connection system for a vertical axis marine current energy converter has been designed and constructed. The grid connection is done with a back-to-back 2L-3L system with a three level cascaded H-bridge converter grid side. The system has been tested in the laboratory and is ready to be installed at the experimental site. Results from the laboratory testing of the system are presented. / Wind Power / Wave Power / Marine Currnet Power