Publikationsserver der Ostbayerischen Technischen Hochschule Regensburg
Not a member yet
6172 research outputs found
Sort by
SiMiC: Context-aware silicon microstructure characterization using attention-based convolutional neural networks for field-emission tip analysis
No full textAccurate characterization of silicon microstructures is essential for advancing microscale fabrication, quality control, and device performance. Traditional analysis using scanning electron microscopy (SEM) often requires labor-intensive, manual evaluation of feature geometry, limiting throughput and reproducibility. In this study, we propose SiMiC: Context-aware Silicon Microstructure Characterization Using Attention-based Convolutional Neural Networks for Field-Emission Tip Analysis. By leveraging deep learning, our approach efficiently extracts morphological features—such as size, shape, and apex curvature—from SEM images, significantly reducing human intervention while improving measurement consistency. A specialized dataset of silicon-based field-emitter tips was developed, and a customized convolutional neural network architecture incorporating attention mechanisms was trained for multiclass microstructure classification and dimensional prediction. Comparative analysis with classical image processing techniques demonstrates that SiMiC achieves high accuracy while maintaining interpretability. The proposed framework establishes a foundation for data-driven microstructure analysis directly linked to field-emission performance, opening avenues for correlating emitter geometry with emission behavior and guiding the design of optimized cold-cathode and SEM electron sources. The related dataset and algorithm repository that could serve as a baseline in this area can be found at https://research.jingjietan.com/?q=SIMIC
Towards Real-World System-Level Integration of Quantum Accelerators: A Hardware/Software Co-Design Approach
No full textThis work-in-progress explores architectural and systemic foundations for integrating quantum accelerators into heterogeneous computing environments. We propose and implement a modular architecture where Quantum Processing Units (QPUs) operate as peripheral devices, supporting pulse-level control interfaces and high-level circuit execution offloading, while internally managing compilation, transpilation, and scheduling. To enable efficient quantum-classical orchestration, we introduce a Quantum Abstraction Layer (QAL) at the operating system level to enable seamless communication, resource management, and integration with existing software frameworks. Our two-step design approach begins with validating the architecture through simulations in virtualised environments. We then implement an FPGA-based surrogate supporting both result- and timing-accurate modes, enabling full-stack emulation and performance evaluation in the absence of physical quantum hardware. This platform supports extensible and rapid prototyping, Hardware/Software Co-Design, and allows for investigations on the practical quantum advantage under realistic system-level constraints of various use cases. We aim for applicability by hardware vendors, facilitating early development even before physical quantum processors are available
Arbeitsengagement und Zufriedenheit in der HAW-Professur – Impulse für eine nachhaltige Personalentwicklung
No full textHintergrund
Während Rekrutierungsaktivitäten für HAW-Professuren sowohl wissenschaftlich (z.B. Braun & Wilson, 2023) als auch im Zuge von Kampagnen (z.B. „Werden Sie Prof!“; „Die HAW-Professur“) in den Fokus rücken, findet die Personalentwicklung von bereits etablierten HAW-Professorinnen und -Professoren noch wenig Beachtung. Dies liegt auch daran, dass die fachliche und berufliche Weiterentwicklung oft in der Eigenverantwortung der Professorinnen/Professoren gesehen wird (Becker, 2020). Dabei können bis zur Pensionierung gut 25 Jahre im Beruf der HAW-Professur verbracht werden (durchschnittliches Alter bei Erstberufung nach eigener Erhebung: 40 Jahre). Über diesen Zeitraum gilt es, Arbeitsengagement und berufliche Zufriedenheit zu bewahren. Aus diesem Grund stellt sich die Frage, ob und wie eine Personalentwicklung wirken kann, um Leistung und Engagement nachhaltig aufrechtzuerhalten.
Forschungsfrage
Die HAW-Professur setzt eine ausgeprägte berufliche Kompetenz (Praxis-, Forschungs- und Lehrerfahrung) voraus und ermöglicht ein hohes Maß an Autonomie (Wilkesmann & Lauer, 2021). Sie bietet daher optimale Voraussetzungen, um psychologische Grundbedürfnisse im Sinne der Self-Determination Theory (Deci & Ryan, 2000) zu erfüllen, welche wiederum ein erfolgreiches Wirken in der Arbeit ermöglichen. Im Rahmen einer deutschlandweiten Befragung von HAW-Professorinnen und -Professoren wurde deshalb der Frage nachgegangen, inwieweit Autonomie- und Kompetenzwahrnehmung und das Zugehörigkeitsgefühl zur Hochschulgemeinschaft das Arbeitsengagement und die Zufriedenheit in der HAW-Professur fördern.
Methodisches Vorgehen und Datengrundlage
Über 700 HAW-Professorinnen und Professoren beteiligten sich an einer Online-Befragung zum Thema „Karrierewege und Selbstbestimmung“. Dabei wurden u.a. das Arbeitsengagement (Bakker & Schaufeli, 2015) und die Zufriedenheit (Bérubé et al.,2007) erfasst. In Anlehnung an die Self-Determination Theory (Deci & Ryan, 2000) wurden außerdem die Aspekte Autonomie- und Kompetenzwahrnehmung sowie soziale Eingebundenheit erhoben.
Ergebnisse
Die Studie liefert Hinweise, wie die Personalentwicklung für HAW-Professuren gestaltet werden kann, um Zufriedenheit und ein hohes Arbeitsengagement aufrechtzuerhalten. Sie lassen also Rückschlüsse auf die Gestaltung einer zielgerichteten und nachhaltigen Personalentwicklung zu.
Diese Forschung wird mit Mitteln des Bundesministeriums für Bildung und Forschung und des Landes Bayern als Teil des Projekts Zukunft akademisches Personal der OTH Regensburg (ZAP.OTHR) unter dem Programm „FH-Personal“ gefördert
Fabrication of Bisphenol-free Boron-Siloxane Polymers
No full textInterest in the preparation of magnetoactive boron-siloxane polymers (MBP) has been stimulated by recent scientific developments. However, recent regulations have prohibited the inclusion of Bisphenol-A normally employed in the manufacture of commercially available products.
In the following publication, two different routes for the preparation of magnetoactive borosiloxane polymers have been taken. First, the possibilities of obtaining homogeneous mixtures of magnetoactive polymers by mixing the magnetic particles with the polymer using solid CO2 (dry ice) and stirring (shearing) are investigated. In the course of the work, an approach for the preparation of reference materials of different viscosities was made to prove the transferability to other elastomers. This is then followed by a brief review of the reaction of the polymer with different solvents in relation to the mixing with particles. In manually kneaded compounds, dispersions ranging from 0.00004 to 0.004 particles per mm3 are common. The process described in this work resulted in agglomerations of no greater than 0.00004 particles per mm3
A possible synthesis route for bisphenol-free boron-siloxanes is then considered, in which for the polymer matrix, polydimethylsilanes (dichloro(methyl)silane), boric acid and iron(III)chloride are used for this purpose by polycondensation
Digitale Zwillinge für die Steuerung innerbetrieblicher Transportsysteme
No full textIm Rahmen des Forschungsprojektes "Digitale(r) Zwillinge zur dynamischen Simulation für die Planung und Steuerung innerbetrieblicher Transportsysteme im Rahmen der digitalen Fabrik" (TwInTraSys) wurde ein simulationsbasierter Digitaler Zwilling zur vorausschauenden Entscheidungsunterstützung im operativen Betrieb von innerbetrieblichen Transportsystemen entwickelt. Das Konzept wurde demonstratorisch umgesetzt und bei verschiedenen Anwendungsunternehmen getestet und validiert. Im Folgenden werden die Einsatzmöglichkeiten im Zentrallager der Firma HiPP am Beispiel einer Ressourceneinsatzplanung dargestellt
Einfluss der Probengeometrie auf die Mechanischen Eigenschaften von Stampflehm - Eine Literaturrecherche
No full textDer Klimawandel lenkt den Fokus im Bauwesen zunehmend auf die Reduktion von Treibhausgasen und die Förderung nachhaltiger Materialien. Baustoffe wie Lehm gewinnen dank ihrer ökologischen und bauphysikalischen Vorteile an Bedeutung. Eine zentrale Herausforderung stellt die Definition standardisierter Prüfmethoden für Stampflehm dar.
Dieser Beitrag analysiert Forschungsergebnisse zur Geometrie von Stampflehmprobekörpern mittels Literaturrecherche und vergleicht experimentelle Untersuchungen sowie relevante Normen und Testmethoden. Die Auswertung zeigt, dass äußere Einflüsse wie Mischungszusammensetzung, Verdichtungsenergie, Gesteinskörnung und Tonminerale eine zentrale Rolle spielen. Eine Vereinheitlichung von Probengeometrie, Lagerung und Prüfverfahren ist essenziell, um reproduzierbare Ergebnisse zu erzielen und die mechanischen Kennwerte von Stampflehm vergleichbar zu machen. Ziel ist es, einheitliche Standards zu entwickeln, um die Stampflehmbauweise in der Praxis zu fördern
Influence of fictitious bristle parameters in dynamic friction models
No full textIn this study, the influence of bristle parameters in dynamic friction models is investigated, in particular in the LuGre and FrD2 models. These models contain internal states to better capture the friction behavior. The FrD2 model, a second-order dynamic friction model, aims at higher accuracy. The investigation focuses on how the independent variation of the bristle stiffness and the damping parameters affects the model behavior, especially considering the practical approximations used in determining the damping coefficient
Controlled Diversity: Length-optimized Natural Language Generation
No full textLLMs are not generally able to adjust the length of their outputs based on strict length requirements, a capability that would improve their usefulness in applications that require adherence to diverse user and system requirements. We present an approach to train LLMs to acquire this capability by augmenting existing data and applying existing fine-tuning techniques, which we compare based on the trained models’ adherence to the length requirement and overall response quality relative to the baseline model. Our results demonstrate that these techniques can be successfully applied to train LLMs to adhere to length requirements, with the trained models generating texts which better align to the length requirements. Our results indicate that our method may change the response quality when using training data that was not generated by the baseline model. This allows simultaneous alignment to another training objective in certain scenarios, but is undesirable otherwise. Training on a dataset containing the model’s own responses eliminates this issue
Mind the naive forecast! a rigorous evaluation of forecasting models for time series with low predictability
No full textIn the field of time series forecasting, numerous machine learning studies have assessed the performance of new methods on highly volatile data from macroeconomics and finance. Unlike in other domains, where models are also compared to simpler statistical or naive baselines, they mostly compare the performance solely relative to other complex models. This approach may lead to limited conclusions and reduce the practical significance of the results, as it overlooks the unpredictability of some highly volatile time series in the datasets used. We apply state-of-the-art methods from time-series econometrics and machine learning, including autoregressive integrated moving average (ARIMA), exponential smoothing (ETS), Bayesian vector autoregressive model (BVAR), long-short term memory neural networks (LSTM), historical consistent neural networks (HCNN), deep vector autoregressive neural networks (DeepVAR), temporal fusion transformers (TFT), and extreme gradient boosting (XGBoost). Our results demonstrate that no method consistently outperforms the naive (no-change) forecast for highly volatile time series from two popular datasets containing exchange rates and stock prices, rendering comparative analysis between complex models less meaningful. In contrast, when applied to more predictable macroeconomic price indices, many of the methods significantly outperform naive forecasts. We find that the performance of machine learning models deteriorates more than that of statistical models for high-volatility time series. This study highlights the critical importance of using appropriate benchmark models, including cost-effective, simple approaches, on datasets that permit meaningful conclusions
Two-stage Approach for Low-dose and Sparse-angle CT Reconstruction using Backprojection
No full textThis paper presents a novel two-stage approach for computed tomography (CT) reconstruction, focusing on sparse-angle and low-dose setups to minimize radiation exposure while maintaining high image quality. Two-stage approaches consist of an initial reconstruction followed by a neural network for image refinement. In the initial reconstruction, we apply the backprojection (BP) instead of the traditional filtered backprojection (FBP). This enhances computational speed and offers potential advantages for more complex geometries, such as fan-beam and cone-beam CT. Additionally, BP addresses noise and artifacts in sparse-angle CT by leveraging its inherent noise-smoothing effect, which reduces streaking artifacts common in FBP reconstructions. For the second stage, we fine-tune the DRUNet proposed by Zhang et al. to further improve reconstruction quality. We call our method BP-DRUNet and evaluate its performance on a synthetically generated ellipsoid dataset alongside thewell-established LoDoPaBCT dataset. Our results show that BP-DRUNet produces competetive results in terms of PSNR and SSIM metrics compared to the FBP-based counterpart, FBPDRUNet, and delivers visually competitive results across all tested angular setups