Technische Universität München
  • München, Bayern, Germany
Recent publications
Unexpected faults that occur in the sensors of the single-phase grid-connected converter (SGC) may deteriorate the control performance of the grid current and dc-link voltage, and pose risks to the connected distributed generation system. This article proposes an unknown input sliding mode observer (UI-SMO)-based sensor fault identification and remediation strategy. Therein, an UI-SMO is established that integrates the observations of the grid current and dc-link voltage with the reconstruction of the unknown input (i.e., load condition of the SGC). Such an observer-based analytical redundancy generates the residuals that indicate the abnormalities of the sensors for fault identification. An adaptive fault-tolerant control (AFTC) strategy for the SGC is developed, where an automated control system reconfiguration with unknown input adaptation is seamlessly implemented by substituting the data reported by the faulty sensors with their observed values. Extensive simulation and experimental studies validate the effectiveness of the proposed UI-SMO-based sensor fault identification and AFTC strategy of the SGC.
The dynamic response of tires directly affects the handling stability and ride comfort of a vehicle, which in turn determines the vehicle ride, handling, and driver perception. Modeling and simulation of tire dynamics typically require a reasonable computational cost while ensuring proper accuracy. In this study, a novel theoretical model, the coupled rigid-flexible ring model, is presented to analyze the characteristics of the in-plane dynamic responses of tires on uneven road surfaces. The proposed new method consists of three primary sub-models: an elastic contact algorithm featured by a flexible ring model, rolling/vibration dynamics represented by a rigid ring model, and an internal-force transmission algorithm linking the rigid and flexible ring models. In this way, the proposed new method has the merits of both high accuracy up to 150 Hz and low computing cost. A contact algorithm based on the two-dimensional flexible ring provides a pressure distribution on the tire-road contact patch and the length of the footprint under different vertical loads. The transient dynamic response is then estimated by combining the rigid ring model with the flexible ring model. The accuracy of the contact algorithm and the transient responses are validated against experimental radial stiffness and the over-cleat tests respectively. The results show that the in-plane dynamics of the tire can be predicted well. In addition, this model is extended to the analysis of the low-speed uniformity of tires with geometric defects and is validated experimentally. It indicates that the novel proposed model offers many application scenarios and extension possibilities.
LLC resonant converters have been widely used in DC/DC conversion because of their high efficiency. However, the converters are highly nonlinear, making it hard to use typical linear control methods. This paper proposes a disturbance-observer-based control approach, which achieves better performance compared with the conventional PI controller. First, according to the extended describing function (EDF) method, the small-signal model of the phase-shift LLC resonant converter is established, which is further simplified to a third-order system. Then, the disturbance-observer-based control is presented, which incorporates a disturbance observer (Dob) to estimate the uncertain effects resulting from external disturbances, circuit parameters variation and model mismatch, and compensate for them in a feedforward way. The main advantage of this control structure is that it can improve the stability and dynamic response simultaneously, where the Dob optimizes the dynamic responses and the PI controller is designed to satisfy the stability requirement. Finally, a 600 W simulation model is built in MATLAB/Simulink to verify the proposed control method, whose superiority over the classic PI controller is proved by the simulation results.
High-strength composite materials are receiving increased attention within the aerospace and transportation industries. These materials, although light in weight, still impart high stiffness when compared to conventional structural materials, e.g., aluminum and steel. Fibers and matrix are the basic constituents of composite materials. During high-speed maneuvering of aircraft and high-speed trains, the composite materials are subjected to dynamic loads in changing temperatures. The dynamic behavior of composites strongly depends on the ambient thermal environment. Furthermore, during the in-situ operation, the quantification of real-time dynamic loads is a challenging task. Therefore, the experimental investigation of the dynamic behavior of several carbon-fiber epoxy laminated composite plates at different temperatures, namely 0°C, 25°C, 50°C, 75°C, 100°C, and 125°C, is carried out using operational modal analysis, to identify the modal characteristics of the structure, and the temperature-dependent modal data of the tested composite plates is given in the supplementary data. The temperature-dependent elastic and damping parameters of the carbon–epoxy laminate are estimated using a genetic algorithm-based parameter identification scheme for different sets of modal contribution. A combined experimental and numerical simulation procedure is implemented to estimate deterministic material parameters at different temperatures. To obtain the in-situ material parameters for a given operating frequency range, the modal contribution is selected such that the operating frequency of interest lies within the considered resonance modes. As an example of matrix-dominated elastic parameter, the shear modulus, has been found to degrade significantly with increasing temperature, and shown a strong correlation with temperature.
The difficulty in quantifying the benefit of Structural Health Monitoring (SHM) for decision support is one of the bottlenecks to an extensive adoption of SHM on real-world structures. In this paper, we present a framework for such a quantification of the value of vibration-based SHM, which can be flexibly applied to different use cases. These cover SHM-based decisions at different time scales, from near-real time diagnostics to the prognosis of slowly evolving deterioration processes over the lifetime of a structure. The framework includes an advanced model of the SHM system. It employs a Bayesian filter for the tasks of sequential joint deterioration state-parameter estimation and structural reliability updating, using continuously identified modal and intermittent visual inspection data. It also includes a realistic model of the inspection and maintenance decisions throughout the structural life-cycle. On this basis, the Value of SHM is quantified by the difference in expected total life-cycle costs with and without the SHM. We investigate the framework through application on a numerical model of a two-span bridge system, subjected to gradual and shock deterioration, as well as to changing environmental conditions, over its lifetime. The results show that this framework can be used as an a-priori decision support tool to inform the decision on whether or not to install a vibration-based SHM system on a structure, for a wide range of SHM use cases.
We consider a notion of origin for deterministic macro tree transducers with look-ahead which records for each output node, the corresponding input node for which a rule-application generated that output node. With respect to this natural notion, we show that “origin equivalence” is decidable — whenever the transducers are weakly self-nesting. The latter means that whenever two nested calls on the same input node occur, then there must be at least one other node (a terminal output node or a call on another input node) in between these nested calls. Besides origin equivalence we are also able to decide “origin injectivity” for such transducers.
In this paper we consider a measure-theoretical formulation of the training of NeurODEs in the form of a mean-field optimal control with L2-regularization of the control. We derive first order optimality conditions for the NeurODE training problem in the form of a mean-field maximum principle, and show that it admits a unique control solution, which is Lipschitz continuous in time. As a consequence of this uniqueness property, the mean-field maximum principle also provides a strong quantitative generalization error for finite sample approximations, yielding a rigorous justification of a phenomenon that we call coupled descent, indicating the simultaneous decrease of generalization and training errors. We consider two approaches to the derivation of the mean-field maximum principle, including one that is based on a generalized Lagrange multiplier theorem on convex sets of spaces of measures, which is arguably much simpler than those currently available in the literature for mean-field optimal control problems. The latter is also new, and can be considered as a result of independent interest.
pyFBS is an open-source Python package for frequency-based substructuring. The package implements an object-oriented approach for dynamic substructuring. This tutorial is intended to introduce structural dynamics and NVH engineers to the research toolbox in order to overcome vibration challenges in the future. The focus will be on experimental modeling and post-processing of datasets in the context of dynamic substructuring applications. The state-of-the-art methods of frequency-based substructuring, such as the virtual point transformation, the singular vector transformation, and system-equivalent model mixing, are available in pyFBS and will be presented. Furthermore, basic and application examples, as well as numerical and experimental datasets that are provided, are intended to familiarize users with the workflow of the package. pyFBS is demonstrated with two example structures. First, a simple beam-like structure is used to demonstrate how to start with the experimental modeling, FRF synthesis, virtual point transformation, and mixing of system equivalence models. Second, an automotive test structure is used to demonstrate the use of the pyFBS on a complex structure where in-situ transfer path analysis is used to characterize the blocked forces. This tutorial is intended to provide an informal overview of how research can be powered by open-source tools.
Lightweight design for gears is becoming increasingly important for efficient, sustainable drive trains. Innovative lightweight designs can be achieved by the additive manufacturing process of P.B.F.-L.B./M. (powder bed fusion by laser beam of metals). This contribution presents lightweight hub designs for gears manufactured by P.B.F.-L.B./M. Helical as well as spur gears are 3D-printed out of the case-hardening steel 16MnCr5. After the PBF-LB/M-process, the gears are case-carburized, shot blasted for mechanical cleaning. The gears with lightweight hubs are analyzed concerning their density, microstructure, roughness. The gears are tested regarding their static, dynamic load carrying capacity, the influence of the lightweight hub on the load carrying capacity is analyzed, evaluated. In conclusion, this contribution enables aprofound understanding, prospective evolution of lightweight hub designs for gears.
Shared vehicles architectures for fuel cell and battery electric vehicles offer a high potential for cost reduction by enabling economies of scale in engineering and production. The efficient integration of hydrogen storages in flat box-shaped battery design spaces represents one of the essential basic requirements. As state-of-the-art cylindrical pressure vessels do not allow a high volumetric efficiency in the installation space, two concepts of box-shaped pressure vessels with tension struts are investigated with regard to manufacturability. The first concept focuses on the integration of aramid fibers in a carbon fiber tank by tufting. In a second concept 3D weaving is analyzed with regard to the construction of a pressure vessel with inner tension struts. For both tank designs manufacturing technologies are developed and the concepts are validated using prototypes. Considering technologies for series production of the textile sector possible paths for industrialization are identified.
Nuclei instance segmentation and classification in histology plays a major role in routine pathology image examination, which enable morphological features analysis that further facilitates streamlined diagnosis and prognosis quantification. However, the nuclei in the tissue images obtained from different human organs are characterized with high variability in shape, size, morphology and spatial arrangements. Moreover, during digitization of tissue slide, the image quality is degraded because of added artifacts, poor contrast, blurred regions due to failed auto-focus and inconsistent staining procedure. Owing to these challenges, it is difficult to build a generalized feature representation that can achieve precise segmentation and classification of nuclei instances in complex tumor micro-environment of tissue specimens obtained from various organs. To address these problems, we propose a novel deep learning model, that harnesses horizontal and vertical distance information hidden among the nuclei instances to successfully delineate the challenging nuclei. Our proposed methodology uses soft attention mechanism to generate relevant feature activation and prune irrelevant and noisy information. These attention units produce more precise and refined feature maps resulting in finer instances segmentation and accurate classification in the overlapping nuclei, the nuclei with touching boundaries and reduction in false positives. We train our model on publicly available data-sets (Kumar, CoNSep, CPM-17 and a new data-set PanNuke). Our methodology shows superior performance in nuclei classification and segmentation in comparison with recently published methods. The code and the obtained results have been made public at the following link:
The increasingly aged human population (mainly in developed countries) represents a significant scientific achievement and privilege associated with medical, social, and economic progress. However, it also poses several challenges to national health and social care systems. The uncoupling of biological evolution with the vast and fast technical progress achieved by humanity has minimized the role of natural selection and rendered aging almost an undesirable physiological event that most people desire to delay as much as possible. All this has been challenging modern gerontology to focus on potential strategies to extend the lifespan, but primarily to mitigate the negative thoughts often associated with aging and aged individuals.
α-Diimine Ni and Pd complexes are one of the most examined late-transition organometallics in the application of catalyzed ethylene (co)polymerization. These organometallic catalysts provide unique advantages and particular opportunities to tailor the architectures, composition, and topology of synthesized polymers through catalyzed polymerization. Two decades after their initial discovery, they are still drawing extensive attention in both academia and industry. More recently, researchers have studied the effect of structural modifications on the α-diimine Ni and Pd complexes and their catalytic behaviors in ethylene (co)polymerization. This review highlights the recent progress in the developments of α-diimine Ni and Pd complexes achieved in the last decade. The chain-walking mechanism as a unique catalytic behavior of α-diimine Ni and Pd complexes is also addressed. The versatile synthesis of ligands and complexes enables researchers to tailor the catalytic performance and the microstructures of polyethylene. Correlations between their structural tunes and catalytic behaviors, polymer properties, and the ethylene copolymerization with polar monomers are comparatively presented and discussed. The heterogenization study of α-diimine Ni and Pd complexes on a solid support for heterogeneous catalysis is also comprehensively summarized. The review is broadly classified into four sections which includes i) the coordination-insertion chemistry in ethylene (co)polymerization, ii) the modification of ligand structure, iii) their application in the field of heterogeneous polymerization, iv) and the properties of the synthesized polymers, followed by the short summary and outlook for their potential studies and applications.
Deep neural networks enable highly accurate image segmentation but require large amounts of manually annotated data for supervised training. Few-shot learning aims to overcome this weakness by learning a new class from a few annotated support samples. This chapter introduces our novel few-shot segmentation framework for volumetric medical images with only a few annotated slices. Compared to other related works in computer vision, the main challenges are the absence of pretrained networks and the volumetric nature of medical scans. We address these challenges by proposing a new architecture for few-shot segmentation that incorporates ‘squeeze & excite’ blocks. Our two-armed architecture consists of a conditioner arm, which processes the annotated support input and generates a task-specific representation. This representation is passed on to the segmenter arm that uses this information to segment the new query image. To facilitate efficient interaction between the conditioner and the segmenter arm, we propose to use ‘channel squeeze & spatial excitation’ blocks – a lightweight computational module – that enables heavy interaction between both the arms with negligible increase in model complexity. This contribution allows us to perform image segmentation without relying on a pretrained model, which generally is unavailable for medical scans. Furthermore, we propose an efficient strategy for volumetric segmentation by optimally pairing a few slices of the support volume to all the slices of the query volume. We perform experiments for organ segmentation on whole-body contrast-enhanced CT scans from the Visceral Dataset. Our proposed model outperforms multiple baselines and existing approaches in segmentation accuracy by a significant margin. The source code is available at
Meta learning or learning to learn has been an attractive topic of research in the past years. Different methods in this area have been proposed to solve existing problems in the machine learning world. One of the common problems in machine learning that has received much attention in recent years is few-shot learning. Meta learning has been the natural solution to many few-shot learning problems. In this chapter, we introduce some background in meta learning. Then, we provide some examples of its applications in different areas, especially in medical imaging.
Cities across the Unites States have embraced green infrastructure (GI) in official planning efforts. The plans conceptualize GI as providing multiple functions and benefits for urban residents, and form part of complex responses to intersectional urban challenges of social injustice and inequity, climate change, aging and expensive infrastructure, and socio-economic change. To date, it is unclear whether official city GI programs address systemic racism and urban inequality. To fill this knowledge gap, we coded and analyzed 122 formal plans from 20 US cities to examine if and how they address equity and justice in three domains: visions, processes, and distributions. We find a widespread failure of plans to conceptualize and operationalize equity planning principles. Only 13% of plans define equity or justice. Only 30% of cities recognize that they are on Native land. Over 90% of plans do not utilize inclusive processes to plan, design, implement, or evaluate GI, and so target many communities for green improvements without their consent. Although 80% of plans use GI to manage hazards and provide multiple benefits with GI, less than 10% identify the causes of uneven distributions and vulnerability. Even fewer recognize related issues of houselessness and gentrification. Very few plans have mechanisms to build community wealth through new GI jobs. We find promising seeds of best practices in some cities and plan types, but no plan exemplified best practices across all equity dimensions. If formal GI planning in US cities does not explicitly and comprehensively address equity concerns, it may reproduce the inequalities that GI is meant to alleviate. Based on our results, we identify-three key needs to improve current GI planning practices for green infrastructure and equity. First, clear definitions of equity and justice are needed, second, planning must engage with causes of inequality and displacement, and third, urban GI planning needs to be transformed through a focus on inclusion.
In this paper, we propose a sequential directional importance sampling (SDIS) method for rare event estimation. SDIS expresses a small failure probability in terms of a sequence of auxiliary failure probabilities, defined by magnifying the input variability. The first probability in the sequence is estimated with Monte Carlo simulation in Cartesian coordinates, and all the subsequent ones are computed with directional importance sampling in polar coordinates. Samples from the directional importance sampling densities used to estimate the intermediate probabilities are drawn in a sequential manner through a resample-move scheme. The latter is conveniently performed in Cartesian coordinates and directional samples are obtained through a suitable transformation. For the move step, we discuss two Markov Chain Monte Carlo (MCMC) algorithms for application in low and high-dimensional problems. Finally, an adaptive choice of the parameters defining the intermediate failure probabilities is proposed and the resulting coefficient of variation of the failure probability estimate is analyzed. The proposed SDIS method is tested on five examples in various problem settings, which demonstrate that the method outperforms existing sequential sampling reliability methods.
Does foreign direct investment (FDI) lead to better or worse labour standards in developing countries? We argue that it depends on the type of labour right, and how costly it is to protect it. We propose that governments are likely to follow international pressure and ‘climb to the top’ of improved labour standards, but only for those rights that do not incur direct costs to foreign investors, such as collective bargaining rights. In contrast, we expect that governments engage in a ’race to the bottom’ when it comes to rights that bear immediate costs for firms, such as overtime pay. To test our argument, we use novel data to distinguish between the legal protection of (1) fair working contracts, (2) adequate working time, (3) dismissal protections, which are more costly; versus (4) collective worker representation, and (5) industrial action rights, which are relatively cheaper to grant. Our panel data analysis for 75 developing countries (1982–2010) shows that higher FDI stock and flow is indeed connected to better protection of collective rights, while FDI flow is connected to a decline in relatively expensive outcome rights. These results remain robust across a range of model specifications.
Purpose Clinical examinations of scoliosis often includes X-rays. Regular clinical monitoring is recommended in particular at young age, because of the high risk of progression during periods of rapid growth. Supplementary methods free of ionizing radiation thus could help to reduce the potential risk of ionizing radiation related health problems. Methods Twelve 3D scan images from female and male patients with different types and severities of spinal deformations were analysed using body scanner image analysis tools. The scan images were captured with a 3D body scanner, which used an infrared sensor and a video camera. To calculate and compare with the patient's specific spinal deformations, simulations based on finite elements methods were performed on biomechanical models of ribcage and spinal column. Results The methods and parameters presented here are in good agreement with corresponding X-rays, used for comparison. High correlation coefficients of ‖ρs‖ ≥ 0.87 between Cobb angle and lateral deviation, as well as between Cobb angle and rotation of the vertebrae, indicate that the parameters could provide supplementary informations in the assessment of spinal deformations. So-called apex angles, in addition introduced to relate the results of the present method with Cobb angles, show strong correlations of ‖ρs‖ ≥ 0.68 and thus could be used for comparison in later follow-up examinations. Conclusion The user-friendly 3D body scanner image analysis tools enable orthopaedic specialists to simulate, visualize and inspect patient's specific spinal deformations. The method is intended to provide supplementary information in complement to the Cobb angle for the assessment of spinal deformations in clinical daily routine and might have the potential to reduce X-rays in follow-up examinations. The Translational Potential of this article The study presents a new method, based on 3D body scanner images and biomechanical modelling, that has the potential to reduce X-rays when monitoring scoliosis especially in young patients.
The crucial role and significance of the electrified electrode/electrolyte interface in optimizing electrochemical systems cannot be underestimated. The net orientation of dipoles and solvent layer structure at the electrode/electrolyte interface can immensely impact electrochemical processes such as the electrode catalytic activity and the charge and mass transfer. The so-called temperature jump effect is refreshingly triggered by employing sub-microsecond laser pulses to irradiate various electrodes. The laser-induced transient techniques are valuable, reliable, and unique tools for determining critical parameters of the electrified interface, such as the potential of maximum entropy (PME) and the potential of zero charge (PZC). Herein, we accentuate the theory behind the techniques and provide relevant information about the experimental setup and design. A detailed summary of recent studies using the laser-induced transient techniques is discussed, with particular emphasis on the relation between the PME/PZC and the electrocatalytic properties of various electrochemical systems.
Institution pages aggregate content on ResearchGate related to an institution. The members listed on this page have self-identified as being affiliated with this institution. Publications listed on this page were identified by our algorithms as relating to this institution. This page was not created or approved by the institution. If you represent an institution and have questions about these pages or wish to report inaccurate content, you can contact us here.
32,955 members
Arne Bethmann
  • SHARE Germany
Arcisstraße 21, 80333, München, Bayern, Germany
Head of institution
Prof. Dr. Thomas Hofmann
+49 89 289 25258
+49 89 289 23399