Gold Open Access- Erstveröffentlichung in einem/als Open-Access-Medium
Refine
Year of publication
Document Type
- Article (341)
- conference proceeding (article) (68)
- Part of a Book (59)
- conference proceeding (presentation, abstract) (7)
- conference proceeding (volume) (7)
- Book (3)
- Edited collection (3)
- Preprint (3)
- Doctoral Thesis (2)
- Working Paper (2)
Is part of the Bibliography
- no (495)
Keywords
- Digitalisierung (8)
- Humans (7)
- Schlaganfall (7)
- magnetoactive elastomer (7)
- Power-to-Gas (6)
- Simulation (6)
- COVID-19 (5)
- Frailty (5)
- Orthogeriatric (5)
- Pflege (5)
Institute
- Fakultät Informatik und Mathematik (137)
- Fakultät Maschinenbau (132)
- Research Center for Artificial Intelligence - RCAI (118)
- Research Center of Health Sciences and Technology - RCHST (106)
- Fakultät Sozial- und Gesundheitswissenschaften (79)
- Fakultät Elektro- und Informationstechnik (77)
- Research Center of Biomedical Engineering - RCBE (66)
- Forschungsstelle für Orthopädie und Ergonomie (31)
- Fakultät Angewandte Natur- und Kulturwissenschaften (30)
- Labor Intelligente Materialien und Strukturen (27)
Begutachtungsstatus
- peer-reviewed (354)
- begutachtet (5)
Model-Driven Software Engineering has long excelled at generating code from static structural models, yet the specification and generation of dynamic behavioral models remains a persistent challenge. Meanwhile, Large Language Models (LLMs) offer flexible, natural-language based code generation but suffer from non-determinism and hallucinations. This paper presents a semi-formal approach that bridges these two paradigms for behavioral modeling via UML state machines. We contribute a textual modeling language that captures the essential elements of UML state diagrams-states, transitions, events, guards, and entry/exit actions-alongside a deterministic code generator that transforms state machine models into Java code following the Gang of Four State design pattern. The language supports two complementary action annotation styles: direct code fragments for concise, self-contained actions, and natural language descriptions for semantically richer behavior to be completed by an LLM weaver . LLM involvement is deliberately scoped to small, well-constrained action bodies, reducing token consumption and non-determinism compared to fully LLM-based approaches. Validated through the Gumball Machine case study, correctness is confirmed by automated tests covering state and transition coverage criteria, and repeating the LLM weaving step produced consistent results across all runs. Compared to both classical UML tooling and fully LLM-based generation, the approach offers stronger determinism, better traceability, lower cognitive modeling effort, and reduced computational cost, while retaining the flexibility to express complex action behavior in natural language where formal specification would be unnecessarily burdensome.
The thermal management-induced drag of conventional ram-air cooling systems for low-temperature fuel-cell propulsion can account for roughly 20% to 25% of total drag in fuel-cell aircraft concepts, while its mass and power impact at the overall aircraft level are far less significant. This drag penalty can severely reduce efficiency, especially when additional parallel power sources for takeoff such as gas turbine engines are undesirable. To address this, we propose augmenting low- and medium-temperature fuel-cell cooling with an auxiliary water-evaporation system. This mechanism is used only when needed, primarily during takeoff in hot ambient conditions, while the ram-air system can be downsized to meet cruise requirements. Water evaporation can achieve coefficients of performance of 50 to 100, while reducing the required mass flow by two orders of magnitude. It provides a heat-rejection energy density of approximately 670 Wh/kg, far exceeding that of state-of-the-art high-power-density batteries. Furthermore, the resulting vapor can be vented overboard, and the system is expected to outperform batteries in reliability, durability, and environmental impact. The paper introduces several architectures for integrating water-evaporation cooling into aircraft systems and discusses their respective advantages, limitations, and implications for overall aircraft performance. Initial results indicate that enabling evaporative cooling can significantly reduce the required ram-air channel size and drag, offering a promising pathway to more efficient fuel-cell-powered aircraft. In the EU project TheMa4HERA, aircraft-level design trades and scaled experimental validation for aviation applications of water evaporation are planned in 2026.
An Extended Simulation-Based Analysis of Car-Sharing Electrification in Schleswig-Holstein, Germany
(2026)
We present a study to assess the feasibility and implications of replacing internal combustion engine vehicles (ICEVs) with battery-powered electric vehicles (EVs) in a car-sharing fleet. For the analysis, we used operational data from a local car-sharing company, which encompasses various aspects such as trip distance, start and duration, vehicle type, and pickup and return locations. To evaluate the impact of transitioning the entire fleet to EVs, we used EV and charger models to simulate the battery-powered trips and the necessary post-trip recharging. Both could affect the service quality of car-sharing services, as the requested trip distance might not be covered by an electric vehicle due to range or charging time limitations. Specifically, in our simulation-based analysis, we identified chains of consecutive bookings as a critical factor for car-sharing electrification. Furthermore, to assess the potential impact of electrification on the energy grid, we used data about the local grid load and its composition to relate it to the predicted vehicle charging times. This is an extended version of our previous paper, incorporating an additional dataset.
Die Digitale Transformation verändert zukünftige Berufsfelder und stellt
somit neue Anforderungen an Studierende. Daher integriert die Regensburg School of Digital Sciences (RSDS) sogenannte »Future Skills« in alle Studiengänge der OTH Regensburg.
Bisherige Studien zu Future Skills konzentrieren sich meist auf die Perspektive von Arbeitgebern oder Lehrenden.Durch offeneDiskussionsrunden und eineOnline-Umfragemit185 Teilnehmernwurden die Future-Skills-Qualifizierungsbedarfe der Studierenden identifiziert.GefragteThemensind »Digitalisierung derArbeitswelt«und »MachineLearning/KI« sowieinterdisziplinäre und projektbasierte Lernräume. Basierend auf diesen Ergebnissen entwickelt die
RSDS neue Kurse, um dieerhobenen Themen in der Lehrezu berücksichtigen.
Diese Arbeit präsentiert ein Konzept zur automatisierten Generierung und Bewertung von Entwurfsvarianten für Linienbaustellen. Basierend auf definierten Rahmenbedingungen und Zielvorgaben werden mittels eines Algorithmus konsistente Simulationsparameter abgeleitet und in ein parametrisches Simulationsmodell überführt. Die Simulation bildet den Bauablauf einer Erdbau-Linienbaustelle ab und ermittelt zentrale Kennwerte wie Bauzeit, Standzeiten, transportierte Materialvolumina und Fahrstrecken. Durch systematische Variation der Einflussgrößen, insbesondere der Fahrzeugflotte, entstehen umfangreiche Datensätze, die sowohl für das Training einer Künstlichen Intelligenz (KI) als auch für die iterative Optimierung verschiedener Bauvarianten eingesetzt werden. Die Ergebnisse zeigen, dass die entwickelte Simulation eine wirksame Optimierung der Transportlogistik ermöglicht. Dabei wird deutlich, dass verschiedene Parameterkonfigurationen zu ähnlichen Ergebnissen führen können und die Gewichtung der Optimierungsziele entscheidend ist. Die Validierung mittels evolutionärem Algorithmus bestätigt die Funktionsfähigkeit des Systems und zeigt, dass eine ausschließlich bauzeitbasierte Bewertung eindimensional ist. Die simulierten Varianten sollen hinsichtlich technischer, wirtschaftlicher und ökologischer Kriterien bewertet werden. Die konkrete Umsetzung erfolgt in Tecnomatix Plant Simulation und bildet die Grundlage für ein automatisiertes Generative-Design-System zur Optimierung von Linienbaustellen.
Die vorliegende Dissertation erörtert das Potenzial der photoakustischen Spektroskopie (PAS) für die Detektion von Methan und Ethan in der Umgebungsluft. Der theoretische Teil widmet sich zunächst ausführlich den Grundlagen der Absorptionsspektroskopie und insbesondere der PAS. Zusätzlich werden für diese Arbeit relevante Messmethoden wie die Absorptionsspektroskopie nach Lambert-Beer, Cavity ring-down Spektroskopie (CRDS) und Wellenlängen-modulierte Spektroskopie (WMS) vorgestellt und ihre jeweiligen Vor- und Nachteile erörtert. Im Theoriekapitel wird zusätzlich die photoakustische Signalgenerierung, einschließlich der Signalverstärkung durch akustische Resonanzverstärkung, detailliert hergeleitet. Besonderes Augenmerk liegt auf der Diskussion des Einflusses der nicht-strahlenden Relaxation auf das photoakustische Signal, wobei die zwei dominanten Formen der stoßbasierten Energieübergänge, Schwingungs-Translation (VT) Relaxation und Schwingungs-Schwingungs (VV) Relaxation, intensiv betrachtet und diskutiert werden. Eine Literaturzusammenfassung rundet den theoretischen Teil ab, wobei der Fokus auf dem Einfluss der Relaxation auf das photoakustische Signal für verschiedene Analyten liegt.
Neben der Charakterisierung der Laserquellen hinsichtlich der Emissionswellenzahl wird ein in die photoakustische Messzelle integriertes System (Acoustic Resonance Monitoring System - ARMS) präsentiert, welches eine schnelle Quantifizierung akustischer Parameter, wie Resonanzfrequenz und Q-Faktor ermöglicht.
Im Ergebnisteil wird der photoakustische Methansensor auf Quereinflüsse gegenüber Sauerstoff, Luftfeuchte, Kohlenstoffdioxid sowie Messzellentemperatur und -druck hin evaluiert. Dabei wurden akustische und relaxationsbedingte Effekte als dominierende Einflussgrößen identifiziert. Die Abhängigkeit der photoakustischen Amplitude und Phase von der Effizienz der nicht-strahlenden Relaxation lässt sich mithilfe des entwickelten Algorithmus CoNRad berechnen und kompensieren, wodurch die Zuverlässigkeit der Sensoren gesteigert wird. Das Konzept des digitalen Zwillings kombiniert theoretische Berechnungen, ARMS-Messungen und die emittierte optische Laserleistung, um ein vollst¨andig theoretisch zu erwartendes photoakustisches Signal zu berechnen und gemessene Rohwerte für sämtliche Quereinflüsse zu kompensieren. Dieser Ansatz wurde in einer mehrtägigen Messreihe mit einem Referenzgerät zum atmosphärischen Methanmonitoring evaluiert. Die Ergebnisse verdeutlichen das Potenzial der PAS zur Spurengasdetektion in der Umgebungsluft, unterstreichen aber auch die Notwendigkeit des digitalen Zwillings zur Quereinflusskompensation.
Messungen mit einem quartz-enhanced PAS Sensor zeigten ebenfalls relaxationsbedingte Signalveränderungen, die mithilfe der etablierten statistischen Methode der Partial Least Squares Regression (PLSR) und dem digitalen Zwilling kompensiert wurden. Mit beiden Ansätzen konnten vergleichbar gute Ergebnisse erzielt werden und den mittleren relativen Fehler der vorhergesagten Analytkonzentration auf den einstelligen Prozentbereich verringert werden.
Darüber hinaus wurde eine Methode entwickelt, um spektrale Quereinflüsse in der Wellenlängen-modulierten PAS zu berechnen. Der Einfluss von spektralen Überlappungen auf das Messsignal wurde zusätzlich für Amplitudenmodulation untersucht.
Zusammenfassend befasst sich diese Dissertation vorrangig mit der detaillierte Analyse von relaxationsbedingten, akustischen und spektralen Einflüssen auf das photoakustische Signal. Hierbei werden verschiedene innovative Ansätze zur Quantifizierung und Kompensation dieser Einflüsse präsentiert, mit dem übergeordneten Ziel, die Genauigkeit und Zuverlässigkeit photoakustischer Sensoren in langfristigen Feldanwendungen signifikant zu erhöhen. Die Erkenntnisse dieser Arbeit unterstreichen nicht nur das Potenzial der photoakustischen Spektroskopie zur präzisen Detektion von Spurengasen in der Umgebungsluft, sondern betonen auch die Notwendigkeit von Kompensationsmechanismen. Insgesamt trägt diese Arbeit entscheidend dazu bei, dass photoakustische Sensoren in kontinuierlichen Feldanwendungen eingesetzt werden können.
This dissertation discusses the potential of photoacoustic spectroscopy (PAS) for the detection of methane and ethane in ambient air. The theoretical section is dedicated to the fundamentals of absorption spectroscopy and PAS in particular. Additionally, spectroscopic measurement methods relevant to this work such as Beer-Lambert absorption spectroscopy, cavity ring-down spectroscopy (CRDS) and wavelength modulated spectroscopy (WMS) are presented and their respective advantages and disadvantages are discussed. The theory chapter also provides a detailed derivation of photoacoustic signal generation, including signal amplification by acoustic resonance amplification. Particular attention is given to the discussion of the influence of non-radiative relaxation on the photoacoustic signal, whereby the two dominant forms of collision-based energy transfer, vibrational-translation (VT) relaxation and vibrational-vibrational (VV) relaxation, are reviewed and discussed. A literature review completes the theoretical part, focusing on the influence of relaxation on the photoacoustic signal for different analytes.
Besides the characterization of the laser sources with respect to the emission wavenumber, an integrated system (Acoustic Resonance Monitoring System - ARMS) is presented, which allows fast quantification of acoustic parameters such as resonance frequency and Q-factor.
The photoacoustic methane sensor is evaluated in the results chapter regarding cross-sensitivities towards oxygen, humidity, carbon dioxide as well as photoacoustic cell temperature and pressure. Acoustic and relaxation-related effects were identified as the dominant parameters. The dependence of the photoacoustic amplitude and phase on the efficiency of the non-radiative relaxation can be calculated and compensated for using the developed algorithm CoNRad, which increases the reliability of the sensor. The concept of the digital twin combines theoretical calculations, ARMS measurements and the emitted optical laser power to calculate a fully theoretically expected photoacoustic signal and to compensate measured raw data for all cross influences. This approach was evaluated in a series of measurements over several days with a reference device for atmospheric methane monitoring. The results illustrate the potential of PAS for trace gas detection in the ambient air, but also emphasize the necessity of the digital twin for signal compensation.
Measurements with a quartz-enhanced PAS sensor also revealed relaxation-related signal distortions, which were compensated using an established statistical method of partial least squares regression (PLSR) and the digital twin. Both approaches achieved equally good results and reduced the mean relative error of the predicted analyte concentration to the single-digit percentage range.
Furthermore, a method for calculating spectral influences in terms of wavelength-modulated PAS was developed. Spectral influences on the measurement signal were also investigated for amplitude modulation.
In summary, this thesis is primarily dedicated to an in-depth analysis of relaxation-related, acoustic and spectral influences on the photoacoustic signal. Various novel approaches to quantify and compensate for these influences are presented, with the objective of significantly increasing the accuracy and reliability of photoacoustic sensors in long-term field applications. The findings of this work not only demonstrate the capability of photoacoustic spectroscopy for the precise detection of trace gases in ambient air, but also emphasize the need for compensation techniques. Overall, this work provides a valuable contribution towards the application of photoacoustic sensors in continuous and long-term field applications.
Temporal regularities and the timing of events and actions such as anticipating enemy movements or planning one’s next move are essential components of almost every video game. Thus, to succeed in video games, it is advantageous to anticipate events and prepare relevant actions before they occur. This work explores whether elapsed time can be used as a predictive cue for implicitly anticipating events in video games. Inspired by findings from psychology, we implemented multiple time-event correlations in a custom video game by pairing specific delays with specific game events. Participants had to shoot targets that appeared at different locations. After a certain delay (e.g., 0.8 s), the targets appeared more frequently (80 % of all appearances) at a specific location (e.g., left up). Our analysis of 25 participants provides evidence that players implicitly learned the implemented time-event correlations and used them to anticipate the location of upcoming targets. This led to improved game performance. Although no participant realised the implemented temporal regularities, targets were shot faster when preceded by the frequently paired delay. Our findings pave the way for game developers and researchers alike to more creatively combine human temporal processing with temporal aspects of video games.
This paper presents an end-to-end workflow for conceptual bridge design that integrates
parametric modeling, automated cost analysis, and data-driven evaluation. At its core lies a top-down
parametric BIM skeleton that governs geometry, earthwork volumes, and adaptive excavation pits.
Through automated IFC exchange and integration with cost-analysis software via an application
programming interface, component volumes are calculated and linked to a template bill of quantities
via matchkeys, ensuring consistent and reproducible cost estimation. The workflow not only enables
reliable analysis of bridge variants based on Key Performance Indicators (KPIs), but also produces
structured synthetic datasets suitable for training machine learning models. In particular, a Conditional Variational Autoencoder is conceptually introduced to demonstrate how such data can support
both forward performance prediction and inverse design conditioned on KPI targets. By improving
data fidelity, the approach reduces the demand for extensive training datasets while increasing the
robustness of model predictions. This research therefore lays the groundwork for bridging parametric
modeling and generative Artificial Intelligence, offering a scalable pathway toward more efficient and
data-driven infrastructure planning.
Soft tensegrity structures with variable stiffness and shape changing abil-ity hold significant potential for soft robotic applications. With few exceptions, this class of structures primarily consists of tensioned and compressed members (at least some of which are compliant) forming a prestressed, stable equilibrium state. Con-sequently, changes in prestress of the structures enable independent or combined adjustments of shape and stiffness, which can be achieved actively or passively. This article explores the fundamental realization principles of soft tensegrity structures focusing on application examples, including manipulators, force and contact sensors, multistable soft grippers, and locomotion systems. Additionally, it highlights the potential of smart materials for passive stiffness modulation. The findings highlight that soft tensegrity structures, characterized by variable stiffness and shape changing capabilities, play a crucial role in advancing the performance and adaptability of soft robotic systems.