05 Fakultät Informatik, Elektrotechnik und Informationstechnik
Permanent URI for this collectionhttps://elib.uni-stuttgart.de/handle/11682/6
Browse
117 results
Search Results
Item Open Access Maschinelles Lernen für intelligente Automatisierungssysteme mit dezentraler Datenhaltung am Anwendungsfall Predictive Maintenance(2019) Maschler, Benjamin; Jazdi, Nasser; Weyrich, MichaelFür eine hohe Ergebnisqualität sind Machine Learning Algorithmen auf eine breite Datenbasis angewiesen. Studien zeigen jedoch, dass viele Unternehmen nicht bereit sind, ihre Daten mit anderen Unternehmen, beispielsweise in Form einer gemeinsamen Daten-Cloud, zu teilen. Ziel sollte es daher sein, effizientes maschinelles Lernen mit einer dezentralen Datenhaltung, die den Verbleib vertraulicher Daten im jeweiligen Ursprungs-Unternehmen ermöglicht, zu ermöglichen. In diesem Artikel wird diesbezüglich ein neuartiges Konzept vorgestellt und hinsichtlich seiner Potentiale für intelligente Automatisierungssysteme am Beispiel des Anwendungsfalls Predictive Maintenance analysiert. Die Umsetzbarkeit des Konzepts unter Nutzung verschiedener bestehender Ansätze wird diskutiert, bevor schließlich auf potentielle Mehrwerte für Anlagenbetreiber sowie -hersteller unter besonderer Berücksichtigung der Perspektive kleiner und mittlerer Unternehmen eingegangen wird.Item Open Access GeSn‐on‐Si avalanche photodiodes with high responsivity and low dark current(2025) Wanitzek, Maurice; Ramachandra, Harishnarayan; Spieth, Christian; Daus, Alwin; Schulze, Jörg; Oehme, MichaelGeSn‐on‐Si avalanche photodiodes (APDs) are emerging as a promising solution for low‐light detection in the short‐wave infrared (SWIR) spectral range, including applications in imaging and telecommunications. In this work, key challenges such as high dark current and limited responsivity are addressed by demonstrating devices, which combine low noise with high signal amplification, while remaining compatible with silicon‐based technology. GeSn‐on‐Si APDs with various Sn concentrations up to 1.9% are fabricated and characterized. The GeSn layers are grown pseudomorphically on Ge virtual substrates on Si wafers using molecular beam epitaxy. The devices comprise a double‐mesa structure and exhibit a dark current dominated by a perimeter leakage path, independent of the Sn content. A dark current below 1 µA is maintained up to the onset of avalanche breakdown, marking a significant improvement compared to prior work. A record‐high responsivity of 14.7 A W -1 is achieved at 1550 nm for the APD with 1.9% Sn. Through impulse response measurements, the 3‐dB bandwidth is determined to 1.2 GHz on devices with an 80 µm diameter, resulting in a responsivity‐bandwidth‐product of 17.6 A W -1 GHz -1 . These results highlight the potential of GeSn‐on‐Si APDs for high‐performance, low‐light applications in the SWIR range.Item Open Access Cycling of double-layered graphite anodes in pouch-cells(2022) Müller, Daniel; Fill, Alexander; Birke, Kai PeterIncremental improvement to the current state-of-the-art lithium-ion technology, for example regarding the physical or electrochemical design, can bridge the gap until the next generation of cells are ready to take Li-ions place. Previously designed two-layered porosity-graded graphite anodes, together with LixNi0.6Mn0.2Co0.2O2 cathodes, were analysed in small pouch-cells with a capacity of around 1 Ah. For comparison, custom-made reference cells with the average properties of two-layered anodes were tested. Ten cells of each type were examined in total. Each cell pair, consisting of one double-layer and one single-layer (reference) cell, underwent the same test procedure. Besides regular charge and discharge cycles, electrochemical impedance spectroscopy, incremental capacity analysis, differential voltage analysis and current-pulse measurement are used to identify the differences in ageing behaviour between the two cell types. The results show similar behaviour and properties at beginning-of-life, but an astonishing improvement in capacity retention for the double-layer cells regardless of the cycling conditions. Additionally, the lifetime of the single-layer cells was strongly influenced by the cycling conditions, and the double-layer cells showed less difference in ageing behaviour.Item Open Access a-Si:H/c-Si heterojunction front- and back contacts for silicon solar cells with p-type base(2010) Rostan, Philipp Johannes; Werner, Jürgen H. (Prof. Dr. rer. nat. habil.)This thesis reports on low temperature amorphous silicon back and front contacts for high-efficiency crystalline silicon solar cells with a p-type base. The back contact uses a sequence of intrinsic amorphous (i-a-Si:H) and boron doped microcrystalline (p-μc-Si:H) silicon layers fabricated by Plasma Enhanced Chemical Vapor Deposition (PECVD) and a magnetron sputtered ZnO:Al layer. The back contact is finished by evaporating Al onto the ZnO:Al and altogether prepared at a maximum temperature of 220 °C. Analysis of the electronic transport of mobile charge carriers at the back contact shows that the two high-efficiency requirements low back contact series resistance and high quality c-Si surface passivation are in strong contradiction to each other, thus difficult to achieve at the same time. The preparation of resistance- and effective lifetime samples allows one to investigate both requirements independently. Analysis of the majority charge carrier transport on complete Al/ZnO:Al/a-Si:H/c-Si back contact structures derives the resistive properties. Measurements of the effective minority carrier lifetime on a-Si:H coated wafers determines the back contact surface passivation quality. Both high-efficiency solar cell requirements together are analyzed in complete photovoltaic devices where the back contact series resistance mainly affects the fill factor and the back contact passivation quality mainly affects the open circuit voltage. The best cell equipped with a diffused emitter with random texture and a full-area a-Si:H/c-Si back contact has an independently confirmed efficiency η = 21.0 % with an open circuit voltage Voc = 681 mV and a fill factor FF = 78.7 % on an area of 1 cm². An alternative concept that uses a simplified a-Si:H layer sequence combined with Al-point contacts yields a confirmed efficiency η = 19.3 % with an open circuit voltage Voc = 655 mV and a fill factor FF = 79.5 % on an area of 2 cm². Analysis of the internal quantum efficiency shows that both types of back contacts lead to effective diffusion lengths in excess of 600 μm. An extended fill factor analysis shows that fill factor limitations for the full-area a-Si:H/c-Si contacts result from non-ideal diode behavior, ascribed to the injection dependence of the heterojunction interface recombination velocity. Analysis of the external quantum efficiency under back side illumination with different bias light intensities delivers the effective surface recombination Seff(Φ) in dependance of the illumination intensity Φ. The front contact (emitter) uses a sequence of intrinsic and phosphorous doped amorphous silicon layers together with a ZnO:Al or a SnO2:In layer and an Al front contact grid. The emitter is prepared at a maximum temperature of 220 °C. Measurements of the minority carrier lifetime on symmetric i/n-a-Si:H coated wafers judge the emitter passivation quality. The best solar cells that use a thermal oxide back side passivation with Al-point contacts and flat a-Si:H emitters have open circuit voltages up to 683 mV and efficiencies up to 17.4 %. The efficiency of such devices is limited by a low short circuit current due to the flat front side. Using the same back contact structure with random pyramid textured wafer front sides and a-Si:H emitters yields open circuit voltages up to 660 mV and efficiencies up to 18.5 %, sofar limited by a relatively low fill factor FF ≤ 74.3 %. Analysis of the external quantum efficiency underlines the excellent surface passivation properties of the amorphous emitter. Combining both, amorphous front- and back contacts yields p-type heterojunction solar cells completely fabricated at temperatures below 220 °C. The best devices reach an open circuit voltage Voc = 678 mV and an efficiency η = 18.1 % with random textured wafers, limited by low fill factors FF ∼ 75 %. Besides the cell fabrication and characterization, this thesis reveals that the inherent a-Si:H/c-Si band offset distribution with a low conduction band offset and a large valence band offset is disadvantageous for p-c-Si heterojuntion solar cells if compared to their n-c-Si counterparts. A calculation of the saturation current densities of the cell's emitter, bulk and back contact demonstrates that the n-a-Si:H/p-c-Si emitter suffers from a low built-in potential. Modelling of the back contact based on the charge carrier transport equations shows that the insertion of an i-a-Si:H layer with a thickness d ≥ 3 nm (that is mandatory for a high surface passivation quality) leads to a series resistance that is critical for usage in a solar cell. The model mainly ascribes the high back contact resistance to the large valence band offset at the heterojunction.Item Open Access Adaptive epitaxy of C‐Si‐Ge‐Sn : customizable bulk and quantum structures(2025) Concepción, Omar; Devaiya, Ambrishkumar J.; Zoellner, Marvin H.; Schubert, Markus A.; Bärwolf, Florian; Seidel, Lukas; Reboud, Vincent; Tiedemann, Andreas T.; Bae, Jin‐Hee; Tchelnokov, Alexei; Zhao, Qing‐Tai; Broderick, Christopher A.; Oehme, Michael; Capellini, Giovanni; Grützmacher, Detlev; Buca, DanThe successful demonstration of (Si)Ge1‐xSnx alloys as direct‐gap materials for infrared lasers has driven intense research on group IV‐based devices for nanoelectronics, energy harvesting, and quantum computing applications. The material palette of direct‐gap group‐IV alloys can be further extended by introducing carbon to fine‐tune their structural and electronic properties, significantly expanding their functionality. This work presents heteroepitaxial growth of C(Si)GeSn alloys using an industry‐standard reduced‐pressure chemical vapor deposition reactor. The introduction of CBr4 as a precursor enables controlled incorporation of C atoms (<1 at.%) into the epilayer lattice, while simultaneously increasing the Sn content in the CGeSn alloy up to ≈18 at.%. Carbon plays a key role in modulating strain, stabilizing the crystal structure, and influencing material properties. By leveraging alloying and strain engineering, quaternary CSiGeSn bulk layers and CGeSn/GeSn heterostructures are epitaxially grown. The impact of C incorporation on optical emission is investigated in LEDs based on CGeSn/GeSn multiple quantum wells, demonstrating enhanced near‐infrared emission at 2.54 µm, which is sustained up to room temperature.Item Open Access Sprachassistierter Entwicklungsprozess für automatisierungstechnische Systeme : ein Ansatz zur Strukturierung komplexer Entwicklungsprozesse(2020) White, Dustin; Weyrich, MichaelDer Systementwicklungsprozess nimmt immer mehr an Komplexität zu, da die Systeme selbst immer komplexer werden. Gleichzeitig Vermischen sich die verschiedenen Disziplinen wie Maschinenbau, Elektrotechnik und Softwaretechnik zunehmend, so dass Unternehmen einer Disziplin sprunghafte Komplexitätszuwächse bei ihren Systemen und in ihrer Entwicklung haben. Deshalb wird in dieser Veröffentlichung ein Konzept eines Sprachassistenten erarbeitet, der durch eine Entwicklungsphase führt. Daraus geht hervor, dass die Software zur Unterstützung der Entwicklung ein Informationsmodell benötigt, um die Daten des entwickelten Systems zu speichern und diese mit dem vorhandenen Wissen zu verbinden. Dieses Wissen kann entweder intern oder im Web vorhanden sein. Der Entwicklungsprozess soll daher Kooperation unterstützen, so dass die Assistenzsoftware und Ingenieure miteinander interagieren.Item Open Access Distributed cooperative deep transfer learning for industrial image recognition(2020) Maschler, Benjamin; Kamm, Simon; Jazdi, Nasser; Weyrich, MichaelIn this paper, a novel light-weight incremental class learning algorithm for live image recognition is presented. It features a dual memory architecture and is capable of learning formerly unknown classes as well as conducting its learning across multiple instances at multiple locations without storing any images. In addition to tests on the ImageNet dataset, a prototype based upon a Raspberry Pi and a webcam is used for further evaluation: The proposed algorithm successfully allows for the performant execution of image classification tasks while learning new classes at several sites simultaneously, thereby enabling its application to various industry use cases, e.g. predictive maintenance or self-optimization.Item Open Access Improving the accuracy of musculotendon models for the simulation of active lengthening(2023) Millard, Matthew; Kempter, Fabian; Stutzig, Norman; Siebert, Tobias; Fehr, JörgVehicle accidents can cause neck injuries which are costly for individuals and society. Safety systems could be designed to reduce the risk of neck injury if it were possible to accurately simulate the tissue-level injuries that later lead to chronic pain. During a crash, reflexes cause the muscles of the neck to be actively lengthened. Although the muscles of the neck are often only mildly injured, the forces developed by the neck’s musculature affect the tissues that are more severely injured. In this work, we compare the forces developed by MAT_156, LS-DYNA’s Hill-type model, and the newly proposed VEXAT muscle model during active lengthening. The results show that Hill-type muscle models underestimate forces developed during active lengthening, while the VEXAT model can more faithfully reproduce experimental measurements.Item Open Access Optimization of disassembly strategies for electric vehicle batteries(2021) Baazouzi, Sabri; Rist, Felix Paul; Weeber, Max; Birke, Kai PeterVarious studies show that electrification, integrated into a circular economy, is crucial to reach sustainable mobility solutions. In this context, the circular use of electric vehicle batteries (EVBs) is particularly relevant because of the resource intensity during manufacturing. After reaching the end-of-life phase, EVBs can be subjected to various circular economy strategies, all of which require the previous disassembly. Today, disassembly is carried out manually and represents a bottleneck process. At the same time, extremely high return volumes have been forecast for the next few years, and manual disassembly is associated with safety risks. That is why automated disassembly is identified as being a key enabler of highly efficient circularity. However, several challenges need to be addressed to ensure secure, economic, and ecological disassembly processes. One of these is ensuring that optimal disassembly strategies are determined, considering the uncertainties during disassembly. This paper introduces our design for an adaptive disassembly planner with an integrated disassembly strategy optimizer. Furthermore, we present our optimization method for obtaining optimal disassembly strategies as a combination of three decisions: (1) the optimal disassembly sequence, (2) the optimal disassembly depth, and (3) the optimal circular economy strategy at the component level. Finally, we apply the proposed method to derive optimal disassembly strategies for one selected battery system for two condition scenarios. The results show that the optimization of disassembly strategies must also be used as a tool in the design phase of battery systems to boost the disassembly automation and thus contribute to achieving profitable circular economy solutions for EVBs.Item Open Access Efficient sampling of transition constraints for motion planning under sliding contacts(2020) Khoury, Marie ThereseIn contact-based motion planning we consider for humanoid and multiped robots problems like going up a staircase, walking over an uneven surface or climbing a steep hill. Solving such tasks requires finding sequences of fixed and sliding contacts and planning the transition from one contact in the environment to another. However, most existing algorithms do not take sliding contacts into account for navigation problems or consider them only for manipulation scenarios. We propose an approach to contact-based planning that uses sliding contacts and exploits contact transitions. Such transitions are elementary operations required for whole contact sequences. To model sliding contacts, we develop a sliding contact constraint that permits the robot to slide on an object’s surface. To exploit contact transitions, we utilize three constraint modes to enable passage: contact with a start surface, no contact and contact with a goal surface. We develop a sampler that samples these transition modes uniformly. In this thesis we focus on the motion of one robot link’s end from an initial contact point toward a designated goal surface while the other end of the robot remains in sliding contact with the initial surface. Our method is evaluated by testing it on manipulator arms of two, three and seven degrees of freedom with different objects and various sampling-based planning algorithms. From the considered manipulator arm, it would be possible to transfer our concept to more complex robots and scenarios and extend it to a whole sequence of contacts.