Kempten University of Applied Sciences

This paper proposes an optimal control strategy for SOC balancing and introduces a framework for analyzing the spatial temperature distribution in a multi‐pack battery energy storage system (BESS) composed of multiple battery modules. While various control techniques exist to distribute power among parallel‐connected battery systems, their influence on the spatial temperature distribution within their modules is often neglected, despite temperature being a critical factor accelerating battery health degradation. To bridge this research gap, this framework integrates a 1D thermal simulation and state‐of‐health (SoH) estimation with power split control strategies. To showcase the application of this framework, a comparative study of two power‐sharing methods is conducted: (i) Model Predictive Control (MPC) based State of Charge (SoC) balancing, and (ii) Rule‐Based Control (RBC) strategies, highlighting their impact on temperature distribution and battery aging. Results show that MPC maintains a more uniform temperature profile, limiting peak temperatures to 300 K and minimizing SoH degradation, whereas RBC results in higher peak temperatures (314 K) and accelerated aging. In summary, this framework primarily intends to: (i) Enable researchers to further develop health‐aware power‐sharing strategies for BESS. (ii) Equip BESS operators with detailed spatial temperature insights to optimize power management and cooling systems.

This paper explores the automation of generating and dispatching Experience API (xAPI) statements for comprehensive tracking of user interactions in e-learning environments. It introduces the react-xapi-wrapper library, an extension of the xAPI JavaScript library designed for use in web applications. Key aspects discussed include the library’s features, its integration into a web-based adaptive learning system (ALS) for software engineering, and the custom verbs used. The goal is to reduce implementation effort for tutors and developers while taking advantage of xAPI’s interoperability, scalability, and ability to track student learning activities and behaviors, laying the foundation for more responsive and personalized learning experiences.

NodeGrade tries to provide a suitable solution for the problem of time-intensive short answer grading. This research focuses simultaneously on performance, functionality and user experience, which is underlined by a triangulated approach. The evaluation results show comparable performance of NodeGrade on public datasets, even outperforming GPT-4 on the SemEval 2013 Task 7. Matching of NodeGrade’s output with multiple human expert raters reveals some weaknesses regarding cases at the lower and upper boundary. In terms of user experience, the interviewed and observed students recognized both positive facets, like better learning support and helpful feedback, and negative sides, including technical limitations and lack of transparency. Overall, NodeGrade promises high potential for further practical use and testing in the field of software engineering education and automatic short answer grading.

Background With the rise of large language models, the application of artificial intelligence in research is expanding, possibly accelerating specific stages of the research processes. This study aims to compare the accuracy, completeness and relevance of chatbot-generated responses against human responses in evidence synthesis as part of a scoping review. Methods We employed a structured survey-based research methodology to analyse and compare responses between two human researchers and four chatbots (ZenoChat, ChatGPT 3.5, ChatGPT 4.0, and ChatFlash) to questions based on a pre-coded sample of 407 articles. These questions were part of an evidence synthesis of a scoping review dealing with digitally supported interaction between healthcare workers. Results The analysis revealed no significant differences in judgments of correctness between answers by chatbots and those given by humans. However, chatbots’ answers were found to recognise the context of the original text better, and they provided more complete, albeit longer, responses. Human responses were less likely to add new content to the original text or include interpretation. Amongst the chatbots, ZenoChat provided the best-rated answers, followed by ChatFlash, with ChatGPT 3.5 and ChatGPT 4.0 tying for third. Correct contextualisation of the answer was positively correlated with completeness and correctness of the answer. Conclusions Chatbots powered by large language models may be a useful tool to accelerate qualitative evidence synthesis. Given the current speed of chatbot development and fine-tuning, the successful applications of chatbots to facilitate research will very likely continue to expand over the coming years.

Light detection and ranging (LiDAR) sensor technology for people detection offers a significant advantage in data protection. However, to design these systems cost- and energy-efficiently, the relationship between the measurement data and final object detection output with deep neural networks (DNNs) has to be elaborated. Therefore, this paper presents augmentation methods to analyze the influence of the distance, resolution, noise, and shading parameters of a LiDAR sensor in real point clouds for people detection. Furthermore, their influence on object detection using DNNs was investigated. A significant reduction in the quality requirements for the point clouds was possible for the measurement setup with only minor degradation on the object list level. The DNNs PointVoxel-Region-based Convolutional Neural Network (PV-RCNN) and Sparsely Embedded Convolutional Detection (SECOND) both only show a reduction in object detection of less than 5% with a reduced resolution of up to 32 factors, for an increase in distance of 4 factors, and with a Gaussian noise up to μ=0 and σ=0.07. In addition, both networks require an unshaded height of approx. 0.5 m from a detected person’s head downwards to ensure good people detection performance without special training for these cases. The results obtained, such as shadowing information, are transferred to a software program to determine the minimum number of sensors and their orientation based on the mounting height of the sensor, the sensor parameters, and the ground area under consideration, both for detection at the point cloud level and object detection level.

Background The demographic transition in Germany is leading to an increase in the number of people needing care or nursing services in their own homes. Interprofessional communication and collaboration among healthcare professions providing outpatient care is paramount to ensure effective and high-quality patient-centred care. However, interprofessional communication and collaboration comes with complex prerequisites and rarely works smoothly. Thus, it is necessary to assess the current status quo. Therefore, the aim is to characterize communication patterns, factors influencing interprofessional communication and collaboration and expectations towards communication and collaboration between home-care nursing services and general practitioner practices in Germany. Methods Semi-structured interviews with healthcare professionals in general practitioners’ practices (n=7) and nurses working in home-care nursing services (n=10) were conducted in southern Germany. The interviews were analysed using inductive thematic content analysis. Results Current communication occurs via fax, telephone or personal contact for various purposes, including issuing or rectifying prescriptions and exchanging information about change in a patient’s condition. Key factors influencing interprofessional communication are organizational (e.g., lack of direct communication), profession-related (e.g., hierarchy) and individual (e.g., capacity to provide care). Interprofessional collaboration is scarce. Healthcare professionals expect uncomplicated, efficient and quick communication and collaboration through set channels. Conclusions Current interaction patterns are deficient and require political, structural and educational changes to establish well-functioning collaboration in the ambulant sector that facilitates patient-centred care. Educational and political reforms should comprise expanding interprofessional education in curricula and the introduction of clear and secure communication channels.

Integrating wire arc additive manufacturing (WAAM) with HEAs presents numerous advantages, notably cost-effectively and efficiently producing large-scale components. However, the successful implementation of WAAM for HEAs necessitates specific filament compositions, which poses challenges. While softer HEAs like Canrtor can be manufactured using solid wire or multicomponent wire cords, fabricating solid wire with the requisite composition for high-hardness alloys becomes unfeasible. Addressing this technological complexity is the focus of this study. The proposed methodology revolves around gas metal arc welding (GMAW), which employs metal powder-cored wires (MPCW). These wires contain powder components in equal proportions, offering advantages over alternative bulk alloy production methods such as vacuum or argon-plasma melting, primarily due to the greater volume of molten material within the workpiece. The refinement of this approach is illustrated using a high-hardness eutectic high-entropy FeCoNiAl alloy system doped with Ta. The resulting WAAMed alloy initially exhibits nearly zero plasticity, a characteristic later mitigated through a specialized heat treatment procedure.

Process curves are multivariate finite time series data coming from manufacturing processes. This paper studies machine learning that detect drifts in process curve datasets. A theoretic framework to synthetically generate process curves in a controlled way is introduced in order to benchmark machine learning algorithms for process drift detection. An evaluation score, called the temporal area under the curve, is introduced, which allows to quantify how well machine learning models unveil curves belonging to drift segments. Finally, a benchmark study comparing popular machine learning approaches on synthetic data generated with the introduced framework is presented that shows that existing algorithms often struggle with datasets containing multiple drift segments.

Flexible working arrangements have become increasingly common and are considered a means to better reconcile paid and unpaid work. Therefore, the use of such measures can determine how couples divide their household and childcare tasks. While currently these tasks are dominantly female connotated, an increase in flexible work arrangements may contribute to a more gender-egalitarian distribution of unpaid work. However, empirical evidence on this association is mixed, and it remains unclear to what extent it differs by gender. Using a sample of 3244 individuals in the German Family Panel of 2018/2019 who were cohabiting with an opposite-sex partner and by applying linear regression models, we tested several hypotheses derived from economic, gender, and time-availability approaches. We separately addressed the division of housework and childcare tasks related to three flexible work measures, namely home-office, schedule flexibility, and working-time autonomy. Contrary to our hypotheses, no flexibility measure seemed to be related to a higher share of household tasks. Rather, any significant association we identified was fully explained through gender: Women took on a larger share of any household task, irrespective of their work flexibility. Only the share of childcare performed seemed to differ by the use of schedule flexibility, as well as by gender. Whereas mothers’ contributions to childcare were larger when they used flexibility, those of fathers were smaller. We conclude that flexible working arrangements do not contribute to a more gender-egalitarian division of unpaid work per se, but the (gendered) motivation to use such flexibility may be decisive.

Technological advancements focus on developing comfortable and acceptable driving characteristics in autonomous vehicles. Present driving functions predominantly possess predefined parameters, and there is no universally accepted driving style for autonomous vehicles. While driving may be technically safe and the likelihood of road accidents is reduced, passengers may still feel insecure due to a mismatch in driving styles between the human and the autonomous system. Incorporating driving style preferences into automated vehicles enhances acceptance, reduces uncertainty, and poses the opportunity to expedite their adoption. Despite the increased research focus on driving styles, there remains a need for comprehensive studies investigating how variations in the driving context impact the assessment of automated driving functions. Therefore, this work evaluates lateral driving style preferences for autonomous vehicles on rural roads, considering different weather and traffic situations. A controlled study was conducted with a variety of German participants utilizing a high-fidelity driving simulator. The participants experienced four different driving styles, including mimicking of their own driving behavior under two weather conditions. A notable preference for a more passive driving style became evident based on statistical analyses of participants’ responses during and after the drives. This study could not confirm the hypothesis that people prefer to be driven by mimicking their own driving behavior. Furthermore, the study illustrated that weather conditions and oncoming traffic substantially influence the perceived comfort during autonomous rides. The gathered dataset is openly accessible at https://www.kaggle.com/datasets/jhaselberger/idcld-subject-study-on-driving-style-preferences .

Frequency-modulated continuous wave (FMCW) radio detection and ranging (RADAR) sensors have become indispensable technologies for automated driving systems (ADS) due to their reliability in adverse weather conditions and their ability to simultaneously measure the distance to objects, relative radial velocity, and azimuth and elevation angles. The automotive industry has increasingly considered simulation-based testing of autonomous vehicles due to safety, cost, and time constraints. This raises the need for virtual environmental perception sensors that provide results close to reality. This work presents the design and structure of a ray-tracing-based, high-fidelity, tool-independent baseband FMCW RADAR sensor model. The RADAR sensor model is developed using the standardized functional mock-up interface (FMI) and open simulation interface (OSI) and is integrated into the co-simulation environment of commercial software to demonstrate its exchangeability. The RADAR FMU model incorporates a multiple input and multiple output (MIMO) 2D linear spacing virtual antenna array, non-coherent integration (NCI) of range-Doppler maps (RDMs) over receiver antennas, a constant false alarm rate (CFAR) to obtain an interim object detection list, and density-based spatial clustering of applications with noise (DBSCAN) to provide a single detection per object. The presented RADAR FMU model also includes RADAR sensor-specific impairments such as phase noise (PN), radio frequency (RF) group delay, phase imbalance (PI) of transmitter antennas, mixer non-linearity including third-order intermodulation products (IM3), and noise figure (NF) of receiver antennas. Additionally, this work presents a methodology for plausibly verifying the RADAR sensor model at the raw data level (range map (RM) and RDM) and object detection list level. The simulation results are compared with real sensor measurements to validate the modeling of sensor-specific impairments. The mean absolute percentage error (MAPE) metric is used to quantify the difference between the simulation and real sensor measurements. The results demonstrate that the complete signal processing toolchain and sensor-specific impairments of the RADAR sensor must be considered to achieve simulation results that closely resemble those of the real sensor.

To make different packages with various filling quantities better comparable for their packaging material use, in this study the packaging material use efficiency was defined as the ratio of fill good amount to the packaging weight. Several hundred rigid packages (tubes, bottles, cans, and carton packages) for liquid and higher-viscous fast-moving consumer goods, e.g., beverages and personal care products, were analyzed (weight) and more than >1000 data sets were taken from packaging suppliers of glass and PET packaging. As expected, glass packaging is heavier than PET packaging by a factor of around 10, and with a higher filling volume less packaging per amount of food is required. The material use efficiency of glass and PET bottles can differ by up to a factor of 3 within one filling quantity. The results are relevant for calculating life cycle assessments (LCAs) and selection of material efficient packaging.