As organizations and critical infrastructure increasingly rely on computer networks for their function, cyber defense becomes more and more important. A recent trend is to employ predictive methods in cybersecurity. Attack projection attempts to predict the next step in an ongoing attack. Previous research has attempted to solve attack projection using deep learning relying solely on LSTM networks. In this work, by contrast, we solved the attack projection problem using three different neural network architectures: an LSTM, a Transformer, and a hybrid LSTM­Transformer model. We then proposed a way to integrate our neural models into an existing software framework that relies on sequential rule mining to predict future security alerts. The models were trained and evaluated on a publicly available dataset of network security alerts and evaluated with respect to precision and recall of alert predictions. We found that the Transformer architecture had the best overall performance in all but one experiment and that the LSTM architecture performed the worst across all experiments. / Då organisationer och kritisk infrastruktur blir alltmer beroende av datornätvärk för sin verksamhet, blir cyberförsvar alltmer viktigt. En pågående trend är att använda prediktiva metoder inom cybersäkerhet. Attackprojicering innebär att försöka förutspå nästa steg i en pågående cyberattack. Tidigare forskning som försökte tillämpa djupinlärning på attackprojicering använde sig enbart av LSTM­nätverk. I detta arbete använde vi däremot tre olika neurala arkitekturer: en LSTM, en Transformer och en LSTM­Transformer­hybrid. Vi föreslog sedan ett sätt att integrera våra modeller med ett befintligt mjukvaruramverk som använder sig av sekventiella regler för att förutspå kommande larm. Modellerna tränades och utvärderades på en publik datamängd och utvärderades med hänsyn till precision och återkallelse. Vi fann att Transformer­modellen hade bäst prestation i alla utom ett experiment och att LSTM­modellen presterade sämst i alla våra experiment.

The resources of mobile networks are expensive and limited, and as demand for mobile data continues to grow, improved resource utilisation is a prioritised issue. Traffic demand at base stations (BSs) vary throughout the day and week, but the capacity remains constant and utilisation could be significantly improved based on precise, robust, and efficient forecasting. This degree project proposes a fully attention- based Transformer model for traffic prediction at mobile network BSs. Similar approaches have shown to be extremely successful in other domains but there seems to be no previous work where a model fully based on the Transformer is applied to predict mobile traffic. The proposed model is evaluated in terms of prediction performance and required time for training by comparison to a recurrent long short- term memory (LSTM) network. The implemented attention- based approach consists of stacked layers of multi- head attention combined with simple feedforward neural network layers. It thus lacks recurrence and was expected to train faster than the LSTM network. Results show that the Transformer model is outperformed by the LSTM in terms of prediction error in all performed experiments when compared after training for an equal number of epochs. The results also show that the Transformer trains roughly twice as fast as the LSTM, and when compared on equal premises in terms of training time, the Transformer predicts with a lower error rate than the LSTM in three out of four evaluated cases. / Efterfrågan av mobildata ökar ständigt och resurserna vid mobila nätverk är både dyra och begränsade. Samtidigt bestäms basstationers kapacitet utifrån hur hög efterfrågan av deras tjänster är när den är som högst, vilket leder till låg utnyttjandegrad av basstationernas resurser när efterfrågan är låg. Genom robust, träffsäker och effektiv prediktion av mobiltrafik kan en lösning där kapaciteten istället följer efterfrågan möjliggöras, vilket skulle minska överflödig resursförbrukning vid låg efterfrågan utan att kompromissa med behovet av hög kapacitet vid hög efterfrågan. Den här studien föreslår en transformermetod, helt baserad på attentionmekanismen, för att prediktera trafik vid basstationer i mobila nätverk. Liknande metoder har visat sig extremt framgångsrika inom andra områden men transformers utan stöd från andra komplexa strukturer tycks vara obeprövade för prediktion av mobiltrafik. För att utvärderas jämförs metoden med ett neuralt nätverk, innefattande noder av typen long short- term memory (LSTM). Jämförelsen genomförs med avseende på träningstid och felprocent vid prediktioner. Transformermodellen består av flera attentionlager staplade i kombination med vanliga feed- forward- lager och den förväntades träna snabbare än LSTM- modellen. Studiens resultat visar att transformermodellen förutspår mobiltrafiken med högre felprocent än LSTM- nätverket när de jämförs efter lika många epoker av träning. Transformermodellen tränas dock knappt dubbelt så snabbt och när modellerna jämförs på lika grunder vad gäller träningstid presterar transformermodellen bättre än LSTM- modellen i tre av fyra utvärderade fall.

Since 2019, COVID-19, as a new acute respiratory disease, has struck the whole world, causing millions of death and threatening the economy, politics, and civilization. Therefore, an accurate prediction of the future spread of COVID-19 becomes crucial in such a situation. In this comparative study, four different time-series analysis models, namely the ARIMA model, the Prophet model, the Long Short-Term Memory (LSTM) model, and the Transformer model, are investigated to determine which has the best performance when predicting the future case trends of COVID-19 in six countries. After obtaining the publicly available COVID-19 case data from Johns Hopkins University Center for Systems Science and Engineering database, we conduct repetitive experiments which exploit the data to predict future trends for all models. The performance is then evaluated by mean squared error (MSE) and mean absolute error (MAE) metrics. The results show that overall the LSTM model has the best performance for all countries that it can achieve extremely low MSE and MAE. The Transformer model has the second-best performance with highly satisfactory results in some countries, and the other models have poorer performance. This project highlights the high accuracy of the LSTM model, which can be used to predict the spread of COVID-19 so that countries can be better prepared and aware when controlling the spread. / Sedan 2019 har COVID-19, som en ny akut andningssjukdom, drabbat hela världen, orsakat miljontals dödsfall och hotat ekonomin, politiken och civilisationen. Därför blir en korrekt förutsägelse av den framtida spridningen av COVID-19 avgörande i en sådan situation. I denna jämförande studie undersöks fyra olika tidsseriemodeller, nämligen ARIMA-modellen, profetmodellen, Long Short-Term Memory (LSTM) -modellen och transformatormodellen, för att avgöra vilken som har bäst prestanda när man förutsäger framtida falltrender av COVID-19 i sex länder. Efter att ha fått offentligt tillgängliga COVID-19-falldata från Johns Hopkins University Center for Systems Science and Engineering-databasen utför vi repetitiva experiment som utnyttjar data för att förutsäga framtida trender för alla modeller. Prestandan utvärderas sedan med medelvärde för kvadratfel (MSE) och medelvärde för absolut fel (MAE). Resultaten visar att LSTM -modellen överlag har den bästa prestandan för alla länder att den kan uppnå extremt låg MSE och MAE. Transformatormodellen har den näst bästa prestandan med mycket tillfredsställande resultat i vissa länder, och de andra modellerna har sämre prestanda. Detta projekt belyser den höga noggrannheten hos LSTM-modellen, som kan användas för att förutsäga spridningen av COVID-19 så att länder kan vara bättre förberedda och medvetna när de kontrollerar spridningen.

No description available.

Abstract. Transformer explosions caused by dielectric failure account for over 50% of the disasters. The aim of this thesis is to examine, compare and outline the differences, in function, as dielectric insulators, vegetables oil has, with respect to the mineral oil used in high-power transformers. We will first consider the vegetable oil which has less dielectric capabilities than the mineral oil used in power transformers. Later in the experiments, we will focus mainly to examine the breakdown voltage property, as we try to alter some properties of the respective oils used. Considering the fact that vegetable oil has low viscosity, with its chemical compounds constituting less molecular masses compared to mineral oil, we endorse, from our experimental findings, that mineral oil is indeed worthy and reasonable to be used as a dielectric in high power transformers. In this write-up, we have considered eleven transformer oil properties. In the experiment proper, we considered only the acidity, whose concentration in the transformer oil increases with aging if the transformer, moisture, and a ‘suitable’ impurity like NaOH(aq). At first glance, one would be tempted to think, as we were, that since the increase in acid content of the oil deteriorates its dielectric performance, an increase in alkaline content of the transformer oil, would increase its dielectric ability; reversing the acid effect. But as we see in the results from our experiments, this is false. We think that the visible degradation of the insulating property of the oil, with the introduction of NaOH(aq), is because it acts as an impurity to suitable dielectric function. From the experiments, the heating procedures resulted in the production of toxic gases. This indicated the actual loss of chemical structure and significant breakage of chemical bonds. The resulting chemical composition of the oil does not produce the same dielectric properties as the initial oil sample. Also, here has been considerable inconsistency in the addition of NaOH(aq) or HCl(aq) to both oils. We only added HCl(aq), before every measurement, in two of the experiments. The other experiments were either with moisture, or a single addition of 2cm3 of either HCl(aq) or NaOH(aq) before heating; after which several measurements were taken, at specific intervals, as the mixture cools. We did so, in the latter, in which we had only one addition of a 2cm3 chemical, because in real life, given the short time frame of the experiment, the total amount of acid in the oil has a negligible change. So, in a functioning heated transformer, within a short time frame, there is actually deterioration in oil insulation properties

This master´s thesis describes the development proposal and production of instrument current transformer in one turn primary winding with 4000//5/5 A transfer. The thesis involves a theoretical analysis, a numerical calculation, a developmental 3D model, corresponding simulations and a standard testing of the transformers. There is a complete written description of measuring current transformer cycle from the development to the production. The transformer development and production works were collaborated with the company KPB Intra s.r.o. The company engages in development, production and sale of these kinds of instrument transformers in the Czech Republic and foreign markets.

<p>U ovoj doktorskoj disertaciji prikazana je koncepcija, realizacija i potvrda nove metode<br />jednovremenog etaloniranja strujnih mernih transformatora sa dve različite merne<br />aparature. Pregledom stručne literature ne postoji podatak da je ova metoda ranije<br />primenjivana. U disertaciji su prikazane različite merne metode za ispitivanje i<br />etaloniranje mernih transformatora koje imaju primenu u savremenoj praksi i koje<br />podrazumevaju i različite merne mogućnosti. Takođe je i ekperimentalno potvrđeno<br />jednovremeno ispitivanje i etaloniranje strujnih mernih transformatora sa nekoliko<br />mernih aparatura koje su zasnovane na različitim metodama. Sprovedna istraživanja i<br />eksperimentalni rezultati pokazuju i potvrđuju niz prednosti ovakvog načina<br />etaloniranja. Detaljna analiza komponenti merne nesigurnosti pokazala je da se<br />primenom ove metode postiže poboljšanje merne nesigurnost etaloniranja za skoro red<br />veličine u odnosu na klasično pojedinačno etaloniranje sa dve različite merne<br />aparature. Analiza uticajnih veličina na mernu nesigurnost pokazuje da se po ovoj novoj<br />metodi etaloniranja eliminiše niz komponenti od kojih su najznačajnije: uticaj<br />nejednakosti referentnih struja i ispitnog opterećenja. Istraživanja su takođe pokazala<br />da jednovremena metoda osim što doprinosi podizanju tačnosti etaloniranja strujnih<br />transformatora, ima primenu i u etaloniranju mernih aparatura za ispitivanje tačnosti<br />mernih transformatora i interkonparaciji strujnih etalon transformatora. Kroz<br />konkretne primere realizovane u praksi, razmotreni su i prikazani načini etaloniranja<br />mernih aparatura za ispitivanje tačnosti ovom novom metodom. Interkomparacijom dva<br />merna sistema visoke klase tačnosti Nacionalnog merološkog instituta Kanade<br />(National Research Council Canada), od kojih je jedna razvijena u Elektrotehničkom<br />Institutu &quot;Nikola Tesla&quot;, pokazana je i prednost primene jednovremene metode u oblasti<br />primarne metrologije strujnih mernih transformatora.</p> / <p>The dissertation presents the concept, its realisation and verification of the new method<br />of simultaneous comparison of the current transformers by two different measuring<br />apparatus. It is shown by searching the literature, that this method has not been used<br />before. In this dissertation different measuring methods for testing and calibration of<br />current transformers, with their different measuring capabilities are presented. Most of<br />them have been used in recent practice. Furthermore, the experimental verification of<br />new simultaneous calibration method is presented. For this reason some measuring<br />apparatus based on different measuring methods were used. Conducted research and<br />experimental results confirmed a number of advantages of this calibration method.<br />Detailed analysis of the components of the uncertainty of measurements shown that<br />using simultaneous method uncertainty of measurements have been improved<br />comparing to method with two individual calibration by different apparatus. In that case<br />some of the measuring uncertainty components can be neglected. The most significant<br />is component caused by variation of referent current and component caused by variation<br />of burden. The research has also showed that simultaneous method can be used for<br />calibration of measuring apparatus for current transformer accuracy testing and their<br />inter-comparisons. The ways of calibration of apparatus for current transformer accuracy<br />testing are considered and presented, through concrete examples realized in practice.<br />A high-accuracy comparison of two NRC (National Research Council Canada)<br />calibration systems were carried out by new simultaneous method. One measuring<br />system is developed at Electrical Engineering institute Nicola Tesla, Belgrade.<br />Accordingly the advantage of simultaneous method applied at the primary metrology of<br />current transformer is verified.</p>

No description available.

Transformatorn har en viktig uppgift för att elsystemet ska fungera optimalt och det är därav väldigt viktigt att den inte går sönder genom att exempelvis överhettas. Från att transformatorn har varit placerad utomhus har det nu blivit allt vanligare att placera den i en omslutande byggnad, vilket påverkar effektiviteten för kylningen av transformatorn. Detta eftersom hastigheten på det passerande luftflödet kring transformatorn blir lägre vilket leder till att temperaturen i luften runtomkring ökar. I detta examensarbete undersöktes lufttemperaturen i en transformatorstation i Västernorrland, i syfte att se hur transformatorn klarar av de belastningar och utomhustemperaturer som den utsätts för. Detta för att kunna säkerställa att temperaturgränser och riktlinjer för interna och externa temperaturer för en transformator uppfylls. Transformatorn som användes i undersökningen har en maximal skenbar effekt på 16 MVA och använder sig av kylsystemet ONAN. Byggnaden runtomkring transformatorn har två ventilationsluckor på nedre långsidan, samt två ventilationsluckor på övre kortsidan.  Målet med undersökningen var att genomföra en teoretisk analys av hur kylningen i den valda transformatorstationen dimensioneras, där simuleringar även skulle göras i syfte att validera den teoretiska analysen. De belastningar som undersökts har utgått ifrån tillhandahållna data ifrån den högsta lasten under en vanlig sommar- och vinterdag. Ett framtida fall har även undersökts där lasten antas gå på märkeffekt under en längre tidsperiod samt under en väldigt varm sommardag, för att se hur hårt transformatorn kan belastas i extrema förhållanden utan att gränser och riktlinjer överskrids. Det framtida fallet har delats upp i två scenarier, extremfall 20 samt extremfall 30, där skillnaden är vilken temperatur in i transformatorstationen de har. Alternativa lösningar för ventilationsluckorna har även studerats, gällande placering på väggar, storlekar samt gallers modell. Matematiska beräkningsmodeller för bland annat luftflödet, stationstemperaturen samt lindningsoch oljetemperaturer utvecklades fram under arbetet gång, vilka samlades i en Excel beräkningsmall. Simuleringar av byggnaden och transformatorn gjordes i COMSOL Multiphysics, där både 2D och 3D modeller undersöktes i syfte att dels analysera värmespridningen i oljan, dels den naturliga ventilationen. Utifrån de matematiska beräkningsmodellerna framgick det att vinterfallet körde på ca 49% belastning, medan sommarfallet körde på ca 10% belastning. Dessa båda fallen klarade alla gränser och riktlinjer kring externa och interna temperaturer för alla areastorlekar, placeringar och gallersmodeller som testades. I extremfallen uppfylldes de interna temperaturökningsgränserna, men extremfall 30 klarade inte den externa temperaturgränsen i något simuleringstest. Skulle ett extremfall 30 i framtiden inträffa, bör fläktar vid radiatorerna eller ventilationsluckorna övervägas, alternativt en större lucköppning där det enligt framräknade resultat behövs en förstoring av öppningarna på 57%. Ytterligare ett alternativ skulle kunna vara att placera ventilationsluckorna i taket, då detta visade sig ge bästa möjliga kylning av transformatorn i simuleringarna. Detta examensarbete skulle kunna användas som en grund inför framtida undersökningar och den framarbetade Excel beräkningsmallen kan användas som riktlinje vid dimensionering av inomhustransformatorstationer. / The transformer plays a crucial role for the electrical system to function optimally, making its reliability vital to prevent issues such as overheating. Traditionally, the transformer has been positioned outdoors. Nowadays it has become increasingly common to house transformers in enclosed buildings, which affects the cooling efficiency of the transformer. This enclosure reduces the speed of airflow around the transformer, subsequently raising the ambient air temperature. In this thesis, the air temperature in a transformer station in Västernorrland was investigated, to assess how the transformer withstands the loads and external temperatures it encounters. This to ensure that requirements and guidelines for internal and external temperatures for the transformer are met. The transformer used in the study has a maximum apparent power of 16 MVA and uses the ONAN cooling system. The enclosing building is equipped with two ventilation hatches on the longer lower side and two on the shorter upper side.  The aim of the investigation was to conduct a theoretical analysis of the cooling system’s dimensions at the selected substation, complemented by simulations to validate the theoretical findings. The loads investigated have been based on the data provided from the highest load during a normal summer and winter day. Additionally, a future scenario was explored where the transformer operates at rated power for extended periods during a very hot summer day to determine the maximum load the transformer can handle under extreme conditions without breaching the set requirements and guidelines. The future case has been divided into two scenarios, extreme case 20 and extreme case 30, where the difference is what temperature into the substation they have. Alternative design solutions for the ventilation hatches have also been studied, regarding placement on walls, sizes, and fire damper model. Mathematical calculation models for, among other things, the air flow, station temperature, winding- and oil temperatures were developed during the project and compiled into an Excel calculation template. Simulations of the building and the transformer were made in COMSOL Multiphysics, analysing both 2D and 3D models with the aim of studying the heat spread in the oil and the natural ventilation.  The mathematical models showed that the winter scenario operated at approximately 49% load, while the summer scenario operated at about 10% load. These two cases passed all requirements and guidelines regarding external and internal temperatures for all tested hatch sizes and locations. In the extreme cases, the internal temperature rise requirement was met. However, extreme 30 failed to meet the external temperature requirement in any simulation test. Should an extreme case 30 occur in the future, fans at the cooling fins or ventilation hatches may be necessary, or potentially enlarging the hatch openings by 57% as suggested by the calculations. Another alternative could be placing the ventilation hatches on the roof, as this arrangement provided optimal cooling in the simulations. This thesis could be used as a basis for future investigations and the developed Excel calculation template can be used as a guideline when dimensioning indoor transformer stations.

Esta pesquisa teve como objetivo o desenvolvimento de um algoritmo computacional capaz de diagnosticar as condições operativas de transformadores de potência. As variáveis tomadas como base para classificar as condições apresentadas se referem às usualmente empregadas pela lógica de proteção diferencial, ou seja, a corrente diferencial e o conteúdo harmônico presente nos sinais em análise. Além disto, foi também verificada a relação entre alguns dos fenômenos associados à falta de qualidade da energia elétrica originados pelos consumidores conectados no secundário e refletidos ao primário do transformador. O algoritmo desenvolvido utilizou-se da Transformada Wavelet e de técnicas de inteligência artificial (Lógica Fuzzy e Redes Neurais Artificiais), com o objetivo de inferir sobre os relacionamentos supracitados. Todos os testes para validação da metodologia proposta foram realizados dispondo de um sistema elétrico simulado no software ATP (Alternative Transients Program). Os resultados encontrados denotam que, frente às condições analisadas, correntes diferenciais e conteúdo harmônico indesejado podem vir a surgir, fazendo com que a lógica implementada venha a diagnosticar erroneamente a condição de operação enfrentada. Os resultados avaliam ainda a propagação destas condições do secundário para o primário do transformador em análise. / This research aimed to develop an algorithm able to diagnose the operating conditions of power transformers. The variables considered to classify the presented conditions were the normally used by differential protection logic, i.e., the differential current and the harmonic content in the signals in analysis. Moreover, the relationship between some phenomena associated to a poor power quality originated from consumers connected to the secondary and reflect to the primary side of the transformer was also verified. In order to infer the relationships above mentioned, the developed algorithm used the Wavelet Transform and artificial intelligence techniques (Fuzzy Logic and Artificial Neural Networks). All the tests applied to validate the proposed methodology were performed making use of the ATP (Alternative Transients Program). In face of the analyzed conditions, the results show that differential currents and undesired harmonic content can arise and the implemented logic will erroneously diagnose the operation condition addressed. The results also illustrate the propagation of these conditions from secondary to primary side of the analyzed transformer.