275003 research outputs found
Sort by
All-indoor optical customer premises equipment for fixed wireless access
No full textWe demonstrate an LED-based optical wireless link for fixed wireless access applications, at data rates of 1.5 Gbit/s over 50 m. Transmission between indoor equipment and outdoor access point is possible through metal-coated insulation glass
Towards Understanding Crossover for Cartesian Genetic Programming
No full text308314Unlike in traditional Genetic Programming, Cartesian Genetic Programming (CGP) does not commonly feature a recombination/crossover operator, although recombination plays an important role in other evolutionary techniques, including Genetic Programming from which CGP originates. Instead, CGP mainly depends on mutation and selection operators in their evolutionary search. To this day, it is still unclear as to why CGP’s performance does not generally improve with the addition of crossover. In this work, we argue that CGP’s positional bias might be a reason for this phenomenon. This bias describes a skewed distribution of active and inactive nodes, which might lead to destructive behaviour of standard recombination operators. We provide a first assessment with preliminary results. No final conclusion to this hypothesis can be drawn yet, as more thorough evaluations must be done first. However, our first results show promising trends and may lay the foundationf or future work
A Biomedical Case Study Showing That Tuning Random Forests Can Fundamentally Change the Interpretation of Supervised Data Structure Exploration Aimed at Knowledge Discovery
No full text544552Knowledge discovery in biomedical data using supervised methods assumes that the data contain structure relevant to the class structure if a classifier can be trained to assign a case to the correct class better than by guessing. In this setting, acceptance or rejection of a scientific hypothesis may depend critically on the ability to classify cases better than randomly, without high classification performance being the primary goal. Random forests are often chosen for knowledge-discovery tasks because they are considered a powerful classifier that does not require sophisticated data transformation or hyperparameter tuning and can be regarded as a reference classifier for tabular numerical data. Here, we report a case where the failure of random forests using the default hyperparameter settings in the standard implementations of R and Python would have led to the rejection of the hypothesis that the data contained structure relevant to the class structure. After tuning the hyperparameters, classification performance increased from 56% to 65% balanced accuracy in R, and from 55% to 67% balanced accuracy in Python. More importantly, the 95% confidence intervals in the tuned versions were to the right of the value of 50% that characterizes guessing-level classification. Thus, tuning provided the desired evidence that the data structure supported the class structure of the data set. In this case, the tuning made more than a quantitative difference in the form of slightly better classification accuracy, but significantly changed the interpretation of the data set. This is especially true when classification performance is low and a small improvement increases the balanced accuracy to over 50% when guessing.2
Investigation of hydrogen generation in rolling and sliding contact with different lubricant formulations Untersuchung der Wasserstoffbildung beim Roll- und Gleitkontakt mit unterschiedlichen Schmierstoffformulierungen
No full text343352Early failure of rolling bearings and the formation of white etching cracks is still a major problem in many technical applications. It is assumed that hydrogen is generated in tribological contact and leads to embrittlement of the rolling bearings, which fail as a result of the changed material properties. To investigate the hydrogen release of different lubricant formulations, an in-situ measurement method for an oscillating sliding test was developed at Fraunhofer IWM. This setup is based on an electrochemical cell, which detects the hydrogen generated in the tribological contact. This in-situ method allows the detection of hydrogen generation in tribological contact under real conditions.2023241
Reducing the Number of CFD Computer Experiments by Use of Latin Hypercube Design and Kriging Regression on the Example of Side Flushing in Sinking EDM
No full text151159The continuous development in the tool and mould making and aero-space industry requires advanced materials which are processed specifically by electrical discharge machining (EDM). The removal of debris from the working gap is of utmost importance to prevent arcing and short circuits, causing higher tool wear and lower material removal rates. Besides passive flushing, manually aligned side flushing nozzles are used in practice to ensure sufficient flushing in deep or large-scale cavities. This paper presents an approach of combining numerical simulations and statistical methods to reduce the number of required computer experiments and calculate optimized side flushing configurations for maximized volume flows. As a result, the number of simulations required to investigate a large range of parameters was reduced by a factor of 106, allowing 108 more combinations to be investigated. The results showed good conformity in terms of increased experimental material removal rate at configurations with increased simulated volume flows.Part F116
Lithography-Based Manufacturing Of Near Net Shape Dispersion Strengthened Copper Parts By Cuprous Oxide Reduction
No full textDispersion strengthening of copper enables the combination of high strength and high conductivity. However, production of dispersion strengthened materials is complex and only yields semi-finished products. The following work presents a new approach based on cuprous oxide as matrix with dispersed alumina. Powder mixtures are printed into a near net shape part via Lithography-based Metal Manufacturing. After debinding, cuprous oxide is reduced to copper in a hydrogen sintering atmosphere. Furthermore, the influence of milling the powder mixtures is investigated with samples produced by film drawing. Characterization includes scanning electron microscopy, Vickers hardness, and electrical conductivity measurements. Electrical conductivity reaches values of up to 78 % IACS. Vickers hardness surpasses pure copper, however, falls short of comparative values from the literature for dispersion-strengthened materials. Reasons for that are insufficient densification and suboptimal dispersoid distribution
Numerical simulation of flake orientation during droplet impact on substrates in spray painting processes
No full textA numerical study of flake orientation by droplet impact on dry and wetted solid surfaces for spray painting processes has been carried out. A dynamic contact angle model was applied for the calculation of viscous droplet impact on the dry surface after experimental validation. A user-defined 6-DOF model that concerns to the rigid body motion was implemented in a commercial CFD program to calculate the flake movement inside the droplet. The simulated flake orientations show interesting results that are helpful to understand and to improve painting processes
Vulnerability-Based Optimal Grid Defense Strategies for Enhancing Cyber-Physical Energy System Resilience
No full textAn approach is proposed to identify optimal asset protection strategies based on vulnerability assessment outcomes. Traditional bilevel attacker-defender models emphasize worst-case scenarios but offer limited defensive guidance. In contrast, trilevel models introduce high computational complexity and rely on fixed network configurations. The proposed critical-components method leverages vulnerability assessment results to determine protection strategies, effectively outsourcing the upper-level defense decision. This enables adaptability to diverse network topologies, assessment techniques, and cyber-physical energy systems without the overhead of multi-level optimization. Case studies demonstrate the potential for improved system resilience across varying operational conditions
Digital simulation environment for wavefront-based alignment process of molded aspheric glass lenses
No full textThe alignment of optical components is one of the fundamental tasks faced in the optical manufacturing industry. If the task is extended to miniaturized aspheric glass lenses, or multi-lens systems, the complexity is further increased. Wavefront sensors are commonly used in the precision alignment of optical components, and they provide an elegant and efficient tool for systems with high complexity. We have implemented an alignment simulation environment which is based on Zemax optics studio (ZOS) and related application interface (API). This tool is used to create a set of linear equations or a neural network which are used to solve the relation between wavefront and optical element alignment. In our first alignment approach, the optical system description is linearized by simulating a set of linear dependencies connecting the known perturbation values and corresponding aberrations in the resulting wavefront described as a set of Zernike coefficients. Based on the simulated perturbation and the resulting Zernike coefficients, the alignment correction action can be determined by solving the resulting set of linear equations in MATLAB. To get more equations to describe shallow dependencies on the on-axis solution, we added options for additional fields and an iterative solution which mimics the real alignment system by minimizing the perturbations through successive corrections in lens position estimation. As a non-linear alternative to computationally expensive equation solving approach, we have also evaluated AI-based neural networks for which the training data was automatically generated in the optical simulator via the ZOS-API. In preliminary evaluations the neural network approach has shown promising performance when compared to the approach with linearized equations
The role of antidrug antibodies in ustekinumab therapy and the impact of methotrexate
No full text39933999Objective: We investigated the impact of concomitant MTX on ustekinumab (UST) levels and antidrug antibody (ADA) formation in PsA and evaluated consequences in pharmacodynamics and pharmacokinetics. Methods: We conducted a post-hoc analysis on 112 PsA serum samples of subjects treated with open-label UST and either concomitant MTX (UST/MTX, n ¼ 58) or placebo (UST/pbo, n ¼ 54) obtained in a randomized (1:1), double-blind, multicentre trial. A validated antibody-binding-based multitiered testing was used to detect ADA and ADA with neutralizing capacity (nADA). The impact of MTX on UST immunogenicity was analysed by comparison of UST/pbo with UST/MTX cohorts at different time points. Patient- and disease-related predispositions for ADA formation were investigated with multiple linear regression analysis. Immunogenicity impact on pharmacokinetics, safety and efficacy was determined by cohort comparison between patients with and without ADA formation. Results: Over 52 weeks, 11 UST/pbo- and 19 UST/MTX-treated patients developed ADA (P > 0.05). In the UST/pbo cohort, the visit-dependent UST levels were in the range of 0.047 (0.05) –0.110 (0.07) mg/ml overall, and 0.037 (0.04)–0.091 (0.08) mg/ml in ADA-confirmed subjects. In UST/ MTX-treated patients, the UST levels exhibited an intervisit variation in the range of 0.0502 (0.04)–0.106 (0.07) mg/ml overall and 0.029 (0.03)–0.097 (0.07) mg/ml in ADA positive subjects (P > 0.05). At week 52, ADA-confirmed patients did not differ significantly (P > 0.05) in safety or clinical outcomes from ADA-negative patients. Conclusion: Concomitant MTX had no significant impact on UST immunogenicity. Furthermore, ADA formation was not associated with impairments in UST safety, efficacy or trough levels.621