Filtern
Erscheinungsjahr
- 2015 (5) (entfernen)
Dokumenttyp
- Beitrag zu einem Tagungsband (4)
- Vortrag (1)
Schlagworte
- Dead reckoning (1)
- Deep learning (1)
- Digital image correlation (DIC) (1)
- Digitale Bildkorrelation (DIC) (1)
- Dynamic testing (1)
- Dynamische Tests (1)
- Indoor positioning (1)
- Inertial tracking (1)
- Machine learning (1)
- Spannungs-Dehnungs-Kurve (1)
Eingeladener Vortrag
- nein (1)
The stress vs. strain curve of materials is affected the rate of imposed straining. Among the methods for dynamic testing the technique known as 'split Hopkinson pressure bar' (SHPB) has evolved into the most widely used one to exert high-speed straining. The theory behind it comprises simple equations to compute stress and strain. The reliability of the strain analysis can be assessed by digital image correlation (DIC). The present results indicate that the visually observed strain is smaller than predicted by theory.
Fiber optic sensors have gained increasing importance in recent years and are well established in many areas of industrial applications. In this paper, we introduce a concept of a self-diagnostic fiber optic sensor. The presented sensor is to resolve the problems of embedded fiber optic sensors in complex structures and to enable the validation under operational conditions. For this purpose, different magnetostrictive coated fiber optic sensors were developed and various experiments were performed to verify their mode of operation and to determine the respective reproducibility. The measuring principle is illustrated by obtained experimental results, which showed a change in wavelength from 1 pm at a magnetic field strength change of 0.25 mT. In addition, the temperature characteristics of the implemented magnetostrictive sensor were analyzed and an experimental factor of 1.5 compared to a reference fiber optic sensor was determined.
In den letzten Jahren ist die Bestimmung der Position eines sich bewegenden Objektes eine immer interessanter werdende Information geworden. Die aktuelle Position oder der zurückgelegte Weg wird mittels miniaturisierter integrierter 3D-Sensoren bestimmt. Durch die Integration dieser Sensoren in moderne Smartphones oder in spezielle robuste Hardware sind die Anwendungsgebiete vielseitig. Die ermittelte Position innerhalb von Gebäuden ist dabei für den Kaufmann innerhalb eines Shoppingcenters aber auch für den Feuerwehrmann im Einsatz wichtig. Für den Kaufmann bedeutet es durch gezielte Angebote potenzielle Kunden anzusprechen und damit seinen Umsatz zu steigern. Für den Einsatzleiter der Feuerwehr ist die genaue Position seiner Feuerwehrmänner im Einsatz lebensrettend und bietet damit eine zusätzliche Sicherheit für die Einsatzkraft. Das hier vorgestellte Multisensorsystem, allgemein als Inertialsystem bezeichnet, und das darauf angewendete Verfahren zur Positionsbestimmung wird in diesem Bericht vorgestellt und näher beschrieben. Die systematischen und zufälligen Messunsicherheitsbeiträge des Inertialsystems und deren Auswirkungen auf die Bestimmung der aktuellen Position und somit auf den resultierenden zurückgelegten Weg werden erörtert.
Location estimation and navigation, especially on smartphones has shown great progress in the past decade due to its low cost and ability to work without additional infrastructure. However, a challenge is the positioning, both in terms of step detection, step length approximation as well as heading estimation, which must be accurate and robust, even when the use of the device is varied in terms of placement or orientation. In this paper, we propose a scheme for retrieving relevant information to detect steps and to estimate the correct step length from raw inertial measurement unit (IMU) data. This approach uses Bidirectional Long Short-Term Memory Recurrent Neural Networks (BLSTM-RNNs). Designed to take contextual information into account, the network can process data gathered from different positions, resulting in a system, which is invariant with respect to transformation and distortions of the input patterns. An experimental evaluation on a dataset produced from 10 individuals demonstrates that this new approach achieves significant improvements over previous attempts and increase the current state-of-the-art results even in the presence of variations and degradations. We achieved a mean classification rate of 98.5% and a standard deviation of 0.70 for 10000 different test sequences and an average error of 1.45% regarding the step length. Thus is the best result on the task gathered in the experiments compared with competing techniques.