05 Fakultät Informatik, Elektrotechnik und Informationstechnik

Permanent URI for this collectionhttps://elib.uni-stuttgart.de/handle/11682/6

Browse

Filter results by typing the first few letters
Now showing 1 - 4 of 4
  • Thumbnail Image
    ItemOpen Access
    Bit-Archäologie - 25 Jahre Computermuseum der Informatik
    (Stuttgart: Computermuseum der Informatik der Universität Stuttgart, 2022) Wiatrowski, Frank; Krause, Klemens; Wiatrowski, Frank (Fotograf); Engstler, Katja Stefanie (Konzept, Redaktion und Gestaltung); Pflüger, Dirk (Verfasser des Grußworts)
    Einblick in die Sammlung des Computermuseums - neun Sammlungsstücke aus der Geschichte der Informatik werden exemplarisch vorgestellt.
  • Thumbnail Image
    ItemOpen Access
    Ensemble approaches for link prediction
    (2024) Braun, Tim
    Knowledge Graphs (KGs) are fundamental for organizing and representing large amounts of information, but they often suffer from incompleteness. Link prediction using Knowledge Graph Embedding (KGE) methods has emerged as a solution to this problem. Many different methods have been proposed to perform link prediction, some of which are a combination of different methods. However, existing approaches that combine different methods typically train models on the entire graph, lacking the diversity seen in machine learning ensembles such as bagging and random forests. In this thesis, we present the novel ensemble approaches UnifEnt and UnifFeat, that divide the KG into sub-graphs by taking advantage of the core principles of bagging and random forests. We evaluated our approach on common KG datasets and showed the benefits of using our method by comparing it to common KGE baseline methods, as well as related work in the area of ensemble methods for link prediction.
  • Thumbnail Image
    ItemOpen Access
    Seamless start-up of a grid-connected photovoltaic system using module-integrated micro-converters
    (2023) Callegaro, Leonardo; Uong, Trung-Hieu; Deilami, Sara
    In traditional grid-tied photovoltaic (PV) installations, when partial shadowing occurs between different PV modules in a string, bypass diodes short-circuit the output terminals of shadowed modules, and the whole system forgoes their potential energy production. This loss can be recovered if a dc-dc converter (micro-converter) is coupled to every PV module, and operated at the maximum power point (MPP). In this scenario, without communication links between the distributed micro-converter and the grid-tied inverter, a start-up procedure must be carefully designed to seamlessly allow the system to transfer PV power to the grid. During this phase, potentially damaging over-voltages and abrupt transients occurring at the micro-converters/inverter interface must be avoided. In this paper, the control algorithm of each micro-converter is enhanced to provide a smooth start-up operation so that PV units can safely start transferring power to the inverter and the grid. Improving from previous works, the proposed control technique is simple and removes the need for current sensors at the output of each micro-converter and at the inverter dc-link, with an economical advantage. Simulation results demonstrate the successful system start-up behavior, whilst confirming the benefits of the proposed control technique. First, the dc-link is energized from the rectified grid voltage. Then, the micro-converters raise the dc-link voltage so that the available PV power is transferred to the grid, with this sequence of operations not causing any abrupt electrical transient. The results also demonstrate the robust behavior of the PV system under non-uniform solar irradiation conditions.
  • Thumbnail Image
    ItemOpen Access
    Unsupervised and generic short-term anticipation of human body motions
    (2020) Enes, Kristina; Errami, Hassan; Wolter, Moritz; Krake, Tim; Eberhardt, Bernhard; Weber, Andreas; Zimmermann, Jörg
    Various neural network based methods are capable of anticipating human body motions from data for a short period of time. What these methods lack are the interpretability and explainability of the network and its results. We propose to use Dynamic Mode Decomposition with delays to represent and anticipate human body motions. Exploring the influence of the number of delays on the reconstruction and prediction of various motion classes, we show that the anticipation errors in our results are comparable to or even better for very short anticipation times (<0.4 s) than a recurrent neural network based method. We perceive our method as a first step towards the interpretability of the results by representing human body motions as linear combinations of previous states and delays. In addition, compared to the neural network based methods large training times are not needed. Actually, our methods do not even regress to any other motions than the one to be anticipated and hence it is of a generic nature.