Multimedia ONline ARchiv CHemnitz
Not a member yet
4978 research outputs found
Sort by
KI-gestützte Topologieoptimierung von nachgiebigen Mechanismen
No full textIn der Produktentwicklung wird Topologieoptimierung oft zur Ermittlung idealer Materialverteilung eingesetzt, jedoch durch hohen Rechenaufwand limitiert. Das IKAT präsentiert einen KI-basierten Ansatz, der diesen Aufwand in die Trainingsphase verlagert und schnelle Geometrieerzeugung ermöglicht. Das vorgestellte Projekt kombiniert KI und TO-basierter Bewertungskriterien, um ein datenunabhängiges KI-Modell zur Generierung nachgiebiger Mechanismen zu entwickeln. Dieses Modell kombiniert einen trainierbaren 'Prädiktor' mit einem bewertenden 'Evaluator“. Erste Implementierungen für Steifigkeitsoptimierung und nachgiebige Mechanismen zeigen Potential. Zukünftig soll die KI in CAD-Programmen zur Ressourcenschonung in der Entwurfsphase eingesetzt werden.In product development, topology optimization is often used to determine ideal material distribution, but it is limited by high computational effort. IKAT presents an AI-based approach that shifts this effort to the training phase and enables rapid geometry generation. The presented project combines AI and TO-based evaluation criteria to develop a data-independent AI model for generating compliant mechanisms. This model combines a trainable 'predictor' with an evaluating 'evaluator.' Initial implementations for stiffness optimization and compliant mechanisms show potential. In the future, the AI is to be used in CAD programs for resource conservation in the design phase
Modellierung eines aktiven Luftfederdämpfers für den Einsatz in einer Gesamtfahrzeugsimulation
No full textLuft-Feder-Dämpfer (nachfolgend als LFD bezeichnet) bieten aufgrund ihrer physikalischen Eigenschaften gegenüber der konventionellen Kombination aus Stahlfeder und hydraulischem Dämpfer funktionale Vorteile für den Einsatz im Fahrwerk eines Fahrzeuges. Die Funktionalität eines LFD lässt sich noch erweitern, indem die Feder- und Dämpfungscharakteristika dynamisch verstellbar gestaltet werden. Abhängig von Aktorgestaltung und -dynamik inkl. der Regelung spricht man dann von adaptiven bis hin zu aktiven Fahrwerken bzw. Komponenten.
In den meisten bisherigen Publikationen wurden verstellbare LFD nur in Hinblick auf Teilaspekte der möglichen Vorteile untersucht, wobei einfache Fahrzeugmodelle wie der Zweimassenschwinger als Abbildung des Viertelfahrzeugs verwendet wurden. Dynamische Anpassungen der Feder- und Dämpfungseigenschaften an einem Rad beeinflussen jedoch in hohem Maße das Verhalten des Gesamtfahrzeuges. Um das Verhalten im Spannungsfeld aus Komfort, Sicherheit und Agilität optimal einzustellen, müssen entsprechende Regelsysteme eine Vielzahl von Ein- und Ausgangsgrößen berücksichtigen. Kernthema des Vortrages ist die Erstellung eines physikalischen LFD-Modells mithilfe von Matlab/Simulink und die anschließende Verwendung dieses Teilmodells in einem Gesamtfahrzeugmodell in der Simulationsumgebung der MKS-basierten Anwendung IPG Carmaker. Die Potenziale eines aktiven LFD werden dabei im Rahmen dieser Co-Simulation in Hinblick auf Fahrkomfort, -stabilität und Agilität mithilfe ausgewählter Fahrmanöver unter realitätsnahen Bedingungen untersucht. Dazu wird ein Regelkonzept zur dynamischen Anpassung des LFD vorgestellt und die erzielten Ergebnisse mithilfe der Visualisierungs- und Auswertungswerkzeuge von IPG Carmaker präsentiert.Air Spring Dampers (from now referred as LFD) offer functional advantages for vehicle chassis applications compared to the conventional combination of steel springs and hydraulic dampers due to their physical properties. The functionality of an LFD can be further enhanced by making the spring and damping characteristics dynamically adjustable. Depending on the actuator design and dynamics, including the control system, this is referred to as adaptive up to active chassis systems and components.
In most previous publications, adjustable LFDs have only been analyzed with regard to specific aspects of their potential benefits, often using simplified vehicle models such as the two-mass oscillator as a representation of a quarter-vehicle model. However, dynamic adjustments of the spring and damping properties at a single wheel significantly influence the behavior of the entire vehicle. To optimally balance the trade-off between comfort, safety, and agility, corresponding control systems must take a variety of input and output variables into account. The main topic of this presentation is the development of a physical LFD model using Matlab/Simulink and its subsequent integration into a full-vehicle model within the simulation environment of the MBS-based application IPG CarMaker. The potential of an active LFD is analyzed within this co-simulation framework concerning driving comfort, stability, and agility, using selected driving maneuvers under realistic conditions. For this purpose, a control concept for the dynamic adjustment of the LFD is introduced, and the obtained results are presented using the visualization and analysis tools of IPG CarMaker
Efficient Fault Location and Characterization in Cable Networks with Known and Unknown Topologies
No full textThis thesis addresses the identification of defects in transmission lines, emphasizing the importance of cable network reliability in critical industries like telecommunications, automotive, and aerospace. Cable faults are classified into hard and soft faults, both of which can lead to significant safety risks and financial losses. Effective fault prediction and localization are essential.
We explore methods like time-domain reflectometry (TDR) and frequency-domain reflectometry (FDR) for fault location, noting their limitations in networks with parallel branches and unknown topology, particularly regarding soft faults.
The thesis introduces three innovative cable diagnosis methods:
1. A cost-effective approach for detecting multiple hard faults in harnesses, using low-cost hardware that detects faults at 1 Hz, outperforming traditional reflectometry techniques.
2. A distributed reflectometry method for bus-shaped networks, employing analytical modeling and an adjacency matrix to resolve reflections and locate faults accurately.
3. A novel TDR algorithm for networks with unknown topology, reconstructing the topology iteratively through reflection analysis using a complex mathematical model. This method significantly enhances processing speed and accuracy.
These methods address various industrial scenarios, ranging from low-risk environments to high-risk industries that require accurate soft fault location, establishing a foundation for future improvements in cable network diagnostics.:1 Introduction
2 Theoretical Background on Transmission Line Theory
3 State of the Art on Cable Fault Identification Techniques
4 Fault Location in Cable Networks with Known Topology
5 Fault Identification in Cable Networks with Unknown Topology
6 Conclusion and Outloo
The emotion-performance relationship in safety-critical systems
No full textThis dissertation examines the emotion-performance relationship and aims to establish a foundation for diagnosing critical and non-critical emotional states. Emotional states such as anger or fear have been demonstrated to influence performance in human-machine systems, potentially resulting in severe negative outcomes, particularly in safety-critical environments. To address this issue, it is essential to establish a formal description of the relationships between user performance and the key parameters valence, arousal, and task demand. For this purpose, a literature review was conducted, and Article 1 proposes the cubic P-EAT model. An analysis of empirical literature indicated that high performance is associated with positive valence, low arousal, and low task demands.
Furthermore, the emotion-performance relationship was examined using a simulated safety-critical environment in an experimental study (N = 50, Article 2). The results revealed interindividual differences: While one cluster of subjects benefited from positive valence and low arousal, a second cluster showed high performance associated with negative valence and low arousal, and a third cluster showed no discernible correlation of emotion and performance. Based on these results, a categorization system of users was proposed, the Affective Response Categories (ARCs). To consider these interindividual differences in safety-critical environments, it was necessary to determine whether they present situational states or stable traits. Article 3 demonstrated overall consistency over time, as was additionally replicated in a second experimental study (N = 17). However, the physiological parameters utilized are complex signals that necessitate cost-intense sensor technology. To design the planned user state diagnosis in a scalable and accessible way, the usage without physiological sensors, relying on self-reported emotions instead, should be feasible. For this purpose, users could be categorized into ARCs based on variations in personality traits. A third experiment (N = 50) showed that the three categories were characterized by differences in the personality traits Neuroticism and Openness to experience (Article 4). A positive valence-performance relationship was associated with higher Neuroticism and lower Openness to experience compared to a negative valence-performance relationship. Based on the results of the present dissertation, a diagnostic component could be developed that first determines ARC membership using a baseline of physiological data or personality self-report. Subsequently, the emotional state could be classified into critical and non-critical using current physiological data or emotional self-report data. For example, states of negative valence would be potentially critical for the cluster characterized by a positive valence-performance relationship. In the future, an affect-adaptive system could optimally support users by addressing critical states with suitable adaptation mechanisms, taking cluster membership into consideration. Thereby, performance decrements could be mitigated or avoided, possibly reducing negative consequences, and increasing safety in high-stakes environments
Design of an inductive sinter module for power packaging: FEM analysis and thermal management characterization based on multi-die applications
No full textIn this work, the thermal behavior of a DBC substrate with a multi-die array using an inductively generated temperature input is investigated. Through a combination of iterative modeling and numerical simulations, the study analyzes the temperature distribution and the resulting temperature profiles within the entire setup. These results provide valuable insights into the practical impact of an alternative temperature generation by induction heating in the field of multi-die attach and power packaging.:1. Introduction
2. Problem description
3. Preliminary results
4. Conclusions
5. References
6. Author
Design of Optimized Broadband Excitation Signals for Stable Impedance Spectroscopy Measurements
No full textElectrochemical impedance spectroscopy requires linearity, causality, and time invariance for reliable measurements, necessitating precise excitation signal design. This thesis presents a methodology for multisine excitation signal optimization through systematic determination of amplitude, frequency distribution, and phase angles. The framework addresses limitations in conventional approaches, particularly for systems undergoing state changes during low-frequency measurements. A novel regularized linear Kramers-Kronig (rLKK) approach evaluates impedance spectra consistency and serves as feedback for signal optimization. Three phase optimization strategies are introduced: an RPG-based metaheuristic that balances stability and linearity constraints, a sigmoid transformation that minimizes crest factor while maintaining linearity, and a two-phase approach that first minimizes state-of-charge variation then optimizes crest factor. Validation on lithium-ion batteries demonstrated quantifiable improvements: 216 ppm deviation and 0.02% overall measurement deviation—10 times lower than published results while reducing acquisition time by factors of 3–9. The proposed stabilization technique for transient behavior further reduced deviations to 84 ppm. When applied to bioimpedance spectroscopy, the methodology accelerated measurement speed by a factor of 19 and improved accuracy by a factor of 2.67 compared to related works while enhancing robustness to noise.:1 Introduction
2 Theoretical Background
3 Overview of Excitation Signals
4 State of the Art of Multisine Signals
5 Design of Optimized Multisine Excitation Signals
6 Comparison of Crest Factor Optimization Methods
7 Experimental Investigations
8 Conclusion
A Evolutionary Role Playing Game Theory
B Regularized Linear Kramers-Kronig (rLKK)
C AC Analysi
Beitrag zur modellbasierten Prozessanalyse des elektrochemischen Präzisionsabtragens von rotationssymmetrischen Außengeometrien
No full textDas Grundprinzip des gepulsten elektrochemischen Abtragens (PECM) mit oszillierender Kathode basiert auf der anodischen Auflösung des metallischen Werkstück-Werkstoffs unter Einwirkung eines externen Stroms. Dies ermöglicht es gratfreie Bauteile, unabhängig der mechanischen Materialeigenschaften ohne Einbringen thermisch oder mechanisch induzierter Spannungen herzustellen. Eine Verfahrensvariante stellt die Formgebung über den umlaufenden inneren Arbeitsabstand beim gepulsten elektrochemischen Abtragen (PECM) mit oszillierender Kathode dar. Im Rahmen dieser Arbeit wurde erfolgreich eine multiphysikalische und mehrskalige Simulationsmethode entwickelt, um sowohl die Feldgrößen im Arbeitsspalt während einer Kathodenoszillation als auch die Formgebung über eine lange Prozesszeit zu berechnen. Auf Basis der Simulationsergebnisse konnte zudem eine geeignete Spülstrategie für die im weiteren Verlauf durchgeführten Experimente festgelegt werden. Unter Anwendung multipler Regressionsmodelle konnte das PECM mit oszillierender Kathode weiterführend charakterisiert und signifikante Einflüsse auf relevante Bearbeitungsergebnisse wie dem Seitenarbeitsabstand identifiziert werden.:1 Einleitung
2 Stand der Wissenschaft und Technik
3 Forschungsbedarf, Zielstellung und Vorgehensweise
4 Experimentelle Realisierung
5 Materialcharakterisierung
6 Simulation des gepulsten elektrochemischen Abtragens mit oszillierender
Kathode
7 Validierung der multiphysikalischen Modelle und Ableitung der Gültigkeitsgrenzen
8 Zusammenfassung der Modellierung und Simulationsergebnisse
9 Charakterisierung des Formgebungsprozesses
10 Zusammenfassung der Ergebnisse und Ausblic
NANODEVICES FOR HYDROGEN-AWARE CONDITION MONITORING TOWARDS PPB DETECTION
No full textWe report on carbon nanotube-based H2 sensors in field-effect transistor configuration fabricated on a wafer-level technology platform. The sensors show a high sensitivity of more than 75 % at 2500 ppm and an extrapolated detection limit towards the sub ppm range. Specifically, we explore a metal nanoparticle functionalization and Poly(4-vinylpyridin) polymeric encapsulation on the sensor`s responsivity and time constant. Parametrical studies unfold the influence of sensor operation regimes by different gate source voltage settings to delineate various operational regimes of the thin-film transistors. There, a thin semi-permeable encapsulation layer with palladium nanoparticles on top enhanced the selectivity of the hydrogen sensor by a factor of 3000 with respect to plain carbon nanotube-based sensors
Influence of the gas supply component delay on the dynamic operation of the PEM fuel cell system
No full textIn mobile applications a dynamic operation of the PEM fuel cell system is needed. This requires a fast adaption of the hydrogen and air mass flow and the respective operating conditions. An insufficient operating strategy can lead to gas starvation, temperature peaks and flooding or drying of the fuel cell which can have negative effects on lifetime and power. One of the main issues is the delayed supply of hydrogen and air to the fuel cell. Within this study the inertia of the air compressor and the hydrogen recirculation unit is analyzed by system simulations revealing possible operating strategies to mitigate critical conditions during dynamic operation. As a basis a fuel cell battery hybrid truck is simulated with respect to a typical driving cycle. The findings of this study will be tested by real driving of the fuel cell truck to validate the developed operating strategy
Model Predictive Temperature Control for Fuel Cell Systems
No full textFuel cells are a promising technology for emission free power generating when fueled with regeneratively produced reactants. During the chemical reaction, heat is produced and is dissipated by a cooling loop in order to assure optimal membrane humidification and thus reliability, efficiency and longevity of the entire fuel cell. In the analyzed setup, a circulation pump pumps the coolant medium through a radiator. A ventilator is used to cool the cooling medium. In this work, a model predictive temperature control is presented, controlling the pump speed and ventilator power to compensate fluctuating heating powers. The developed control is applied on a test bench. Experimental results show tracking of temperature setpoints for the inlet and outlet temperature and constraint satisfaction, while compensating fluctuating heating powers.:1 Introduction
2 Problem Setup
3 Method - Model Predictive Control
4 Results
5 Conclusion
6 Outloo