13979 research outputs found
Sort by
Anwendung von Deep Reinforcement Learning zur Betriebsstrategieoptimierung von Energieversorgungsanlagen der ETA-Fabrik
Full text linkZur Erreichung internationaler Klimaschutzziele ist es unerlässlich, den industriellen Energieverbrauch zu reduzieren und den Übergang zu erneuerbaren Energiequellen zu vollziehen. Die Anwendung von Deep Reinforcement Learning zur Betriebsstrategieoptimierung von Energieversorgungsanlagen bietet Potenziale zur Steigerung der Energieeffizienz und -flexibilität sowie zur Kostensenkung. In der
vorliegenden Arbeit wird daher die testweise reale Anwendung im Energieversorgungssystem der ETA-Fabrik durchgeführt und evaluiert. Dazu wird ein Vorgehen, bestehend aus fünf Schritten gewählt: Definition der Systemgrenzen und Identifizierung der konventionellen Betriebsstrategie, Auswahl des Frameworks und Modellierung des Energieversorgungssystems, Implementierung des Deep Reinforcement Learning Algorithmus, Implementierung des Frameworks und schließlich die reale Anwendung und Evaluierung. Die Kosten des Deep Reinforcement Learning Algorithmus betragen im simulationsbasierten Vergleich 57 % der Kosten der konventionellen Betriebsstrategie. In der testweisen realen Anwendung zeigt der Deep Reinforcement Learning Algorithmus ein sicheres und robustes Betriebsverhalten, was sich durch eine geringe Schalthäufigkeit und das Einhalten der Temperaturgrenzwerte äußert. In nachfolgenden Forschungsarbeiten sind weitere reale Anwendungen sowie die Anpassung der Belohnungsfunktion notwendig, um Vorteile in Bezug auf die Energieeffizienz und -flexibilität zu erreichen. Zusätzlich müssen Verfahren entwickelt werden, um die Kosten im Realsystem unter Berücksichtigung von variierenden Rahmenbedingungen zu quantifizieren
KIVA - Kompetenzentwicklung durch interdisziplinäre Vernetzung von Anfang an
No full textKIVA steht für „Kompetenzentwicklung durch interdisziplinäre Vernetzung
von Anfang an“ und ist ein Projekt zur Förderung von Studium und Lehre an der
TU Darmstadt. Das Vorhaben wird seit Oktober 2011 im Rahmen des „Qualitätspakts Lehre“,
einem gemeinsamen Programm des Bundes und der Länder für bessere Studienbedingungen und mehr Qualität in der Lehre, vom Bundesministerium für Bildung und Forschung gefördert. KIVA möchte für ein Studium der MINT-Fächer begeistern, die interdisziplinäre Vernetzung in der Lehre intensivieren, das studentische Engagement fördern
und die Studieneingangsphase stärken
Automation-aided construction and characterization of Bacillus subtilis PrsA strains for the secretion of amylases
No full textProteins face an obstacle race on their way to successful folding. Chaperones facilitate the proper folding of proteins by ensuring they remain on the correct path toward their final tertiary structure. In bacilli, the PrsA chaperone is essential for the correct folding and stabilization of proteins within the cell wall. Overexpression of the PrsA chaperone has been shown to improve the successful folding and secretion of many biotechnologically relevant secreted enzymes. This resulted in a double benefit: firstly, it promotes the efficient release of properly folded enzymes from the cell wall, and second, it reduces the folding stress for the cell, thereby enhancing the overall fitness of the production organism. This paper presents a workflow in which different wild-type PrsA molecules in Bacillus subtilis are co-expressed with different amylases having different signal peptides and promoters. To achieve this, six genome-reduced strains and nine PrsA proteins were systematically selected based on their cultivation performance and the production of two reference amylases. Following strain selection and deletion of major extracellular proteases, several hundred individual strains were created and screened using a stepwise and modular automation approach combined with amplicon sequencing. In addition to providing the key learnings from the workflow, it was revealed that no single PrsA molecule consistently improved amylase production, but genetic constructs combining different elements showed up to a 10-fold variation in yield. Among the screened constructs, the signal peptides YdjM and YvcE demonstrated the best performance
Synthesis and Investigations of new Modalities for Selective FKBP51 Inhibition
No full textIn this thesis, three different projects are presented: bottom group and cyclohexyl replacements for the SAFit scaffold, macrocyclic SAFit analogs, and a comparative study conducted to elucidate the effect of macrocyclization on physicochemical and selected ADME properties. In the first project, an established and published route was utilized to synthesize alpha-methyl thiophene acetic acid derivatives, which were used to determine the most suitable replacement of cyclohexyl in the SAFit scaffold. Aside from identifying chlorothiophene as the best cyclohexyl replacement, it was discovered that the respective transient binding pocket of FKBP51 is surprisingly compatible with a benzo[b]thiophene moiety. This finding marks a potential starting point for further developments of the SAFit scaffold since it became clear that even larger residues than cyclohexyl are tolerated by the protein. The alpha-methyl group of the chlorothiophene acetic acid was then replaced with different allyl containing moieties using the Ireland-Claisen rearrangement. The resulting double bond was then functionalized with different polar groups. The achieved binding affinities were comparable to the best-known bottom group replacement of Feng et al., but with reduced molecular weight and higher polarity. Unfortunately, further investigations were hampered by the epimerization of the stereocenter of the thiophene acetic acid. The second project focused on SAFit-based macrocycles. First, existing structure-activity relationships (SARs) were completed, and the published findings validated. Then, the smallest macrocyclic scaffold was determined, and two different linkages were investigated. The binding affinities and ADME properties of the amide linkage were so poor that further research was terminated. The 1,4-triazole linkage, investigated together with Vanessa Buffa, resulted in the development of a comprehensive strategy for macrocyclic linker design, to stabilize various conformations. The third project was a comparative study to elucidate the effect of macrocyclization on compounds concerning physicochemical and selected ADME properties. In cooperation with AbbVie Germany and together with Moritz Spiske, 15 macrocycles were selected, and respective matched molecular linear references were synthesized. The results of the measured assay panel revealed that for this series of compounds, macrocyclization significantly improves solubility without impairing the other measured properties
Applying A-PTV to RBC suspension flows
No full textIn strongly confined flow geometries, red blood cells migrate normal to the flow direction, thereby altering the flow rheology of blood. Direct optical measurements can help to gain an improved understanding of these migration processes. In the present study, we demonstrate that astigmatism particle tracking velocimetry is a suitable 3D-particle tracking method that allows to directly measure both 3D concentration and velocity distributions of red blood cells in a flow. Red blood cells assume a non-spherical shape; therefore, the influence of their orientation on the reconstruction of the out-of-plane particle position is evaluated through a ray tracing approach of synthetic, astigmatic images. While for noise-free images, the resulting absolute out-of-plane reconstruction error σz is small for different red blood cell orientations (σz = 0.98 μm), it triples for experimentally relevant signal-to-noise ratios (SNR = 1.2). Reconstruction errors are compared to those of spherical particles. Overall, both the red blood cell orientation and the increase in signal-to-noise ratio induce similar out-of-plane reconstruction error values. Experimental analyses are also performed using both a red blood cell suspension system and a refractive index-matched suspension system of identical volume fraction (1.5%). Comparing results from the red blood cell suspension flow with those of the particulate suspension under identical parameters for volume fraction, particle Reynolds number, and bulk Reynolds number, a similarity in lateral migration behavior is observed under the given conditions. The results indicate that the absolute out-of-plane reconstruction error in the red blood cell suspension system (σz=4.50μm) is approximately 1.5 times larger compared to the refractive index-matched system
Closed-form postbuckling analysis of shear-deformable composite laminated panels
No full textThis paper introduces a closed-form analytical approach to the postbuckling analysis of simply supported shear-deformable composite laminated plates under uniaxial compression. The analysis is based on three different laminate theories in order to explicitly account for transverse shear deformations and uses a geometrically nonlinear formulation in conjunction with the Ritz method in order to enable closed-form analytical expressions for the state variables of buckled composite plates. Results are presented for several different plate configurations, and a comparison is performed with literature results as well as comparative finite element computations which leads to a very satisfying results accuracy. The presented analysis method delivers results without any significant numerical effort and is thus especially suited for practical applications where such postbuckling analyses are performed many times
Auswirkungen der Lehr-Lernüberzeugungen studentischer Physiktutor*innen auf Lehrqualität, Zufriedenheit und Veranstaltungserfolg
No full textVon studentischen Tutor*innen geleitete Übungen sind, insbesondere in der Physik, ein wichtiger Baustein der Hochschullehre. Obwohl dieser Bereich empirisch wenig erforscht ist, legen die bisherigen Forschungsergebnisse nahe, dass die Lehr-Lernüberzeugungen von Tutor*innen ihr Lehrhandeln beeinflussen und dadurch auf die Teilnehmenden wirken. Daher wurde in dieser Studie die folgende Mediationshypothese überprüft:
Die Lehr-Lernüberzeugungen studentischer Physiktutor*innen beeinflussen, vermittelt über ihr Handeln in der Lernunterstützung, die Zufriedenheit und den Veranstaltungserfolg der Teilnehmenden ihrer Übung. Ergänzend wurden mittels Regressionsanalysen die Einflüsse der Lehrerfahrung und des Studienganges der Tutor*innen auf ihre Lehr-Lernüberzeugungen, direkte Effekte dieser Lehr-Lernüberzeugungen auf Zufriedenheit und Veranstaltungserfolg der Teilnehmenden sowie Effekte der Lehrqualität auf Zufriedenheit und Veranstaltungserfolg untersucht.
Zur Erhebung dieser Variablen wurden zunächst Fragebögen entwickelt und pilotiert. Dabei wurde unter anderem auf angepasste Items der COACTIV-Studie zum schulischen Mathematikunterricht zurückgegriffen. Die Lehr-Lernüberzeugungen der Tutor*innen wurden mit zwei Skalen erfasst: eine zur Messung der transmissiven Überzeugungen und eine weitere bezüglich der konstruktivistischen Überzeugungen. Das Lehrhandeln wurde durch die von den Übungsteilnehmenden wahrgenommene Lehrqualität operationalisiert. In der Pilotierung zeigte sich, dass die allgemein-pädagogischen Items aus dem Schulkontext, z. B. bezüglich der Klassenführung, trotz Anpassungen nicht auf die tutorielle Lehre übertragbar waren. Für die fachdidaktische Lehrqualität ergab eine explorative Faktorenanalyse ein Erhebungsinstrument mit den folgenden drei Skalen: Prinzip der minimalen Hilfe, Vergleich von Lösungswegen und fachliches Erklären.
Im Anschluss an die Pilotierung fand im Wintersemester 22/23 und im Sommersemester 2023 die Haupterhebung in verschiedenen Physikbachelorveranstaltungen und Physikveranstaltungen für andere Fachbereiche statt. Erhoben wurde dabei mittels Fragebögen direkt in den Übungsgruppen (Tutor*innen: N = 107; Teilnehmende: N = 793).
In den Regressionsanalysen zeigte sich unter anderem ein sehr starker positiver Zusammenhang zwischen der Qualität der fachlichen Erklärungen und der Zufriedenheit der Teilnehmenden (Varianzaufklärung ca. 55 %). Gleichzeitig wirkte sich häufigeres Vergleichen von Lösungswegen negativ auf die Zufriedenheit und den Veranstaltungserfolg aus. In der Mediationsanalyse konnte ein über eine Reduktion des fachlichen Erklärens vermittelter negativer Effekt der konstruktivistischen Überzeugungen auf die Zufriedenheit der Teilnehmenden gefunden werden. Gleichzeitig gab es einen direkten positiven Einfluss der transmissiven Überzeugungen auf den Veranstaltungserfolg der Teilnehmenden. Diese Ergebnisse widersprechen der theoretischen Erwartung, dass konstruktivistische Überzeugungen positive und transmissive negative Effekte haben.
Unter den gegebenen Rahmenbedingungen legen diese Ergebnisse nahe, das fachliche Erklären der Tutor*innen zu fördern und den Vergleich von Lösungswegen zurückhaltender in den Übungen einzusetzen. Weitere Forschungsvorhaben bezüglich der Lehr-Lernüberzeugungen und des Lehrhandelns von Tutor*innen sind jedoch notwendig
Experimental Studies on Neutron Detection for Nuclear Waste Characterization
No full textProper characterization of radioactive waste has become a pressing challenge for all countries involved in nuclear power production. Often non-destructive analysis techniques are preferred over other methods for safety, economic, and regulatory reasons. As neutrons offer superior penetration capabilities for meter-sized and heavily shielded waste packages while providing good contrast for light materials such as residual water, they have great potential for non-destructive waste analysis. However, a suitable neutron detector is paramount to neutron-based nuclear waste characterization. Therefore, two neutron detectors capable of detecting neutrons spatially and temporally resolved have been evaluated in the work presented here.
The first detector was tailored to the specific requirements of high-energy laser-driven neutron sources as they could potentially be an affordable short-pulsed neutron source in the future. To validate the functionality of this detector based on a scintillator array coupled to a multi-anode photomultiplier tube, two tests were conducted. At the PHELIX laser at GSI, a lead brick was imaged with a contrast of FoM= 1.94, bringing the first proof-of-concept that MeV neutron radiographs can be captured at laser-driven neutron sources with this detector. Furthermore, a more complex T-shaped sample made of plastic (C₃H₄O₂) and tungsten was successfully imaged at the spallation neutron source n_TOF at CERN. Moreover, the temporal resolution of the detector permitted the measurement of energy-dependent transmission spectra of plastic. When comparing the measured data points with expected transmission spectra calculated from the material’s total cross-sections and thickness, reasonable agreement within the uncertainties could be observed.
Additionally, a second neutron detector developed by Adrian Losko has been evaluated. Because this detector is based on an unique event-mode camera utilizing the Timepix3 chip designed at CERN, it offers superior resolution compared to the photomultiplier-based detector. Therefore, a mock-up nuclear waste package comprising, amongst others, light materials such as water and ion exchange resin, has been characterized with the Timepix3-based detector. These representative materials are commonly found in nuclear waste. In addition to capturing accurate radiographs, it was possible to reconstruct the sample in 3D using tomographic reconstruction. From the 3D model, the distance between sample holder table legs was determined to be (4.0 ± 0.1) mm in agreement with the actual distance of (4.0 ± 0.1) mm. The transmission spectra measured with a short-pulse, time-of-flight neutron source enabling energy-resolved neutron detection similarly to laser-driven neutron sources, agreed well with the expected spectra and it has been possible to localize the light materials in the waste package based on their unique transmission profiles
Eric Gobetti: E allora le foibe?
No full textThe author of the short book reviewed here, historian and journalist Eric Gobetti from Turin, has a feel for unusual or provocative titles. For instance, he entitled his book about the Italian occupation of large parts of the Kingdom of Yugoslavia since April 1941 "The cheerful occupation" (L'occupazione allegra). Normally, we would not expect even the worst cynics to refer to the burning down of villages, the establishment of concentration camps, the taking of hostages, and mass executions during the two-year Italian occupation of parts of Yugoslavia as "cheerful." Nor can the failure to mention these war crimes during the peace negotiations between the allied powers and Italy be described as "cheerful" either. In fact, the use of the adjective refers to the subsequent denial of the crimes committed during the Italian occupation, which was reinforced with the help of the myth of the "good Italian," who never harmed anyone but, on the contrary, was always the victim of violence inflicted by others: victims of the German Wehrmacht and Waffen-SS, when, in 1943, the Italian occupation troops in the area from the Greek islands to the Rhȏne Valley were disarmed and taken into captivity and/or subjected to forced labour, and victims of the Yugoslav partisans, who were responsible for killing civilians in Italy’s eastern border regions in September 1943 and again in May 1945
Exploring the evolving landscape of human-centred crisis informatics: current challenges and future trends
Full text linkModern Information and Communication Technology (ICT) has been used in safety-critical situations for over twenty years. Rooted in Human-Computer Interaction (HCI) and related disciplines, the field of crisis informatics made considerable efforts to investigate social media use and role patterns in crises, facilitate the collection, processing and refinement of social media data, design and evaluate supportive ICT, and provide cumulative and longitudinal research. This narrative review examines contemporary challenges of human-centred crisis informatics and envision trends for the following decade, including (I) a broadening scope of crisis informatics, (II) the professionalisation of cross-platform collaboration of citizen communities and emergency services, (III) expert interfaces for explainable and multimodal artificial intelligence for user-generated content assessment, (IV) internet of things and mobile apps for bidirectional communication and warnings in disruption-tolerant networks, as well as (V) digital twins and virtual reality for the effective training of multi-agency collaboration in hybrid hazards