Analytische Chemie
Filtern
Dokumenttyp
- Beitrag zu einem Tagungsband (1)
- Vortrag (1)
Sprache
- Englisch (2)
Referierte Publikation
- nein (2)
Schlagworte
- Damage Detection (2) (entfernen)
Organisationseinheit der BAM
Eingeladener Vortrag
- nein (1)
Understanding on how a machine learning model interprets data is a crucial step to verify its reliability and avoid overfitting. While the focus of the scientific community is nowadays orientated towards deep learning approaches, which are considered as black box approaches, this work presents a toolbox that is based on complementary methods of feature extraction and selection, where the classification decisions of the model are transparent and can be physically interpreted. On the example of guided wave benchmark data from the open guided waves platform, where delamination defects were simulated at multiple positions on a carbon fiber reinforced plastic plate under varying temperature conditions, the authors could identify suitable frequencies for further investigations and experiments. Furthermore, the authors presented a realistic validation scenario which ensures that the machine learning model learns global damage characteristics rather than position specific characteristics.
Temperature compensation strategies play a key role in the implementation of guided wave based structural health monitoring approaches. The varying temperature influences the performance of the inspection system inducing false alarms or missed detection, with a consequent reduction of reliability. This paper quantitatively assesses two temperature compensation methods, namely the optimal baseline selection (OBS) and the baseline signal stretch (BSS), with the aim to extend their use to the case of distributed sensor networks (DSN). The effect of temperature separation between baseline time-traces in OBS and BSS are investigated considering multiple couples of sensors employed in the DSN. A decision strategy that uses frequent value warning to define the optimal baseline or stretching parameter is found to be effective analyzing data from two several experiments, which use different frequency analysis with either predominantly A0 mode or S0 mode data or both. The focus is given on the fact that different paths are available in a sensor network and several possible combinations of results are available. Nonetheless, introducing a frequent value warning it is possible to increase the efficiency of the OBS and BSS approach making use of fewer signal processing algorithms. In addition, the effectiveness of those approach is quantified using damage indicators as metric, which confirms that the performance of OBS and BSS quantitatively agree with predictions and also demonstrate that the use of compensation strategies improve detectability of damage with a higher reliability of the system.