Refine
Year of publication
- 2020 (66) (remove)
Document type
- Conference Proceeding (66) (remove)
Has full text
- No (66)
Keywords
- Industry 4.0 (5)
- Machine learning (5)
- Poster (3)
- Assembly line (2)
- Blockchain (2)
- Force amplification mechanisms (2)
- Human factors (2)
- Human-robot collaboration (2)
- Manufacturing (2)
- Piezoelectric actuator (2)
In this paper, the influence of current sensors of a NILM system is investigated. The current sensors of a classical inductive current transformer and a Rogowski coil are compared. To evaluate the actual influence on the NILM, measurements are performed with two measuring systems with different current sensors. With these measuring systems, 20 different consumers with 50 switch-on and switch-off cycles are measured in parallel. Besides, the influence of the sampling rate on the results of the NILM classification is evaluated. The classification is carried out with features normalized to the performance and without phase information, so only the signal waveform is used to differentiate the devices.
Machine learning applications, like machine condition monitoring, predictive maintenance, and others, become a state of the art in Industry 4.0. One of many machine learning algorithms are decision trees for the decision-making process. A new approach for creating distributed decision trees, called node based parallelization, is presented. It allows data to be classified through a network of industrial devices. Each industrial device is responsible for a single classification rule. Also, nodes that react incorrectly, for example, due to an attack, are taken into account using a variety of methods to remain the decision-making process correct and robust.
Corrosion is a process that needs to be viewed carefully in context with the examined metals or alloys as well as the ambient conditions (e.g. electrolyte composition). Additive manufacturing processes with their formation a of microscale melt and rapid solidification of that melt can lead to microstructures that can differ extremely from conventional manufacturing processes in terms of their homogeneity and distribution of (alloying) elements. Therefore, process–related local inclusions can be formed with higher amounts of certain alloying elements than their surroundings which result in different chemical potentials. Corrosion experiments performed with additive manufactured parts (e.g. made of pure titanium or titanium alloys) show the release of potentially unwanted alloy constituents, which in turn can affect the long–term behavior of the part negatively. As part of the investigations it is shown what kind of influence the additive manufacturing process can have on such built parts and how subsequently applied treatments like machining or heat treatment can alter the properties of the material and produced component. Different methods like metallography or potentiodynamic polarization with subsequent mass spectrometric analyses were eventually performed to investigate the mentioned material properties and behaviour.
Additive Manufacturing is a highly innovative and pioneering process that offers among others a high degree of flexibility and complexity in terms of the part design or the possibility to integrate various functions in a single part. Therefore, it possesses great chances to establish itself as a significant method within the entire field of manufacturing processes in the near future. The used materials and their thermodynamic behavior determine the resulting properties of parts built in this way, but also by the generated microstructure. Regarding the whole process with its formation of a microscale melt and ongoing rapid solidification a variety of different microstructures can be created, which in turn can affect the mechanical as well as chemical properties and the long–term behavior to a great extent. Furthermore, it can be seen that different metals and alloys in combination with the process conditions can result in different and/or fluctuating qualities of the manufactured components. Nonetheless, additive manufacturing can lead to a noticeably enhancement of materials or products that were manufactured and processed with traditional methods so far and open new possibilities and perspectives in the research and development sector. However, this means that it is crucial to adapt currently used tests and methods to the new properties and manufacturing process.
EIT Based Time Constant Analysis to Determine Different Types of Patients in COVID-19 Pneumonia
(2020)
Validation of Continuously Learning AI/ML Systems in Medical Devices – A Scenario-based Analysis
(2020)
Investigation of Long-Term Stability of Hybrid Systems-in-Foil (HySiF) for Biomedical Applications
(2020)
Schwimmen in Plastik
(2020)
Das humane Mikrobiom
(2020)
Optimization of Microfabricated 2D Planar Spiral Microcoils for the Micro NDT of Grinding Burn
(2020)
Swimming in Plastic, A Story
(2020)
In situ Qualitätsbeurteilung von Schleifprozessen mittels Mikrosystemtechnik basierter Sensorfusion
(2020)
We introduce an algorithm that performs road background segmentation on video material from pedestrian perspective using machine learning methods. As there are no annotated data sets providing training data for machine learning, we develop a method that automatically extracts road respectively background blocks from the first frames of a sequence by analyzing weights based on mean gray value, mean saturation, and y coordinate of the block’s middle pixel. For each block labeled either road or background, several feature vectors are computed by considering smaller overlapping blocks within each block. Together with the x coordinate of a block’s middle pixel, mean gray value, mean saturation, and y coordinate form a block’s feature vector. All feature vectors and their labels are passed to a machine learning method. The resulting model is then applied to the remaining frames of the video sequence in order to separate road and background. In tests, the accuracy of the training data passed to the machine learning methods was 99.84 %. For the complete algorithm, we reached hit rates of 99.41 % when using a support vector machine and 99.87 % when using a neural network.