Filtern
Erscheinungsjahr
Dokumenttyp
- Zeitschriftenartikel (14)
- Beitrag zu einem Tagungsband (8)
- Vortrag (6)
- Beitrag zu einem Sammelband (1)
- Dissertation (1)
Schlagworte
- Guided waves (30) (entfernen)
Organisationseinheit der BAM
Eingeladener Vortrag
- nein (6)
The applications of composite overwrapped pressure vessels (COPVs) in extreme conditions, such as storing hydrogen gases at very high pressure, impose new requirements related to the system's integrity and safety. The development of a structural health monitoring (SHM) system that allows for continuous monitoring of the COPVs provides rich information about the structural integrity of the component. Furthermore, the collected data can be used for different purposes such as increasing the periodic inspection intervals, providing a remaining lifetime prognosis, and also ensuring optimal operating conditions. Ultimately this information can be complementary to the development of the envisioned digital twin of the monitored COPVs. Guided waves (GWs) are preferred to be used in continuous SHM given their ability to travel in complex structures for long distances. However, obtained GW signals are complex and require advanced processing techniques. Machine learning (ML) is increasingly utilized as the main part of the processing pipeline to automatically detect anomalies in the system's integrity. Hence, in this study, we are scrutinizing the potential of using ML to provide continuous monitoring of COPVs based on ultrasonic GW data. Data is collected from a network of sensors consisting of fifteen Piezoelectric (PZT) wafers that were surface mounted on the COPV. Two ML algorithms are used in the automated evaluation procedure (i) a long short-term memory (LSTM) autoencoder for anomaly detection (defects/impact), and (ii) a convolutional neural network (CNN) model for feature extraction and classification of the artificial damage sizes and locations. Additional data augmentation steps are introduced such as modification and addition of random noise to original signals to enhance the model's robustness to uncertainties. Overall, it was shown that the ML algorithms used were able to detect and classify the simulated damage with high accuracy.
The applications of composite overwrapped pressure vessels (COPVs) in extreme conditions, such as storing hydrogen gases at very high pressure, impose new requirements related to the system's integrity and safety. The development of a structural health monitoring (SHM) system that allows for continuous monitoring of the COPVs provides rich information about the structural integrity of the component. Furthermore, the collected data can be used for different purposes such as increasing the periodic inspection intervals, providing a remaining lifetime prognosis, and also ensuring optimal operating conditions. Ultimately this information can be complementary to the development of the envisioned digital twin of the monitored COPVs. Guided waves (GWs) are preferred to be used in continuous SHM given their ability to travel in complex structures for long distances. However, obtained GW signals are complex and require advanced processing techniques. Machine learning (ML) is increasingly utilized as the main part of the processing pipeline to automatically detect anomalies in the system's integrity. Hence, in this study, we are scrutinizing the potential of using ML to provide continuous monitoring of COPVs based on ultrasonic GW data. Data is collected from a network of sensors consisting of fifteen Piezoelectric (PZT) wafers that were surface mounted on the COPV. Two ML algorithms are used in the automated evaluation procedure (i) a long short-term memory (LSTM) autoencoder for anomaly detection (defects/impact), and (ii) a convolutional neural network (CNN) model for feature extraction and classification of the artificial damage sizes and locations. Additional data augmentation steps are introduced such as modification and addition of random noise to original signals to enhance the model's robustness to uncertainties. Overall, it was shown that the ML algorithms used were able to detect and classify the simulated damage with high accuracy.
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.
Um aus messtechnisch ermittelten Dispersionsabbildungen geführter Ultraschallwellen Rückschlüsse auf die Materialparameter zu ziehen, werden in der aktuellen Forschung verschiedene inverse Methoden diskutiert. Maschinelles Lernen und insbesondere Convolutional-Neural-Networks (CNNs) stellen eine Möglichkeit der automatisierten inversen Modellierung und Evaluierung von Bilddaten dar. In diesem Vortrag wird anhand synthetischer Daten gezeigt, wie das Ausbreitungsverhalten von geführten Ultraschallwellen unter Verwendung von CNNs genutzt werden kann, um die isotropen elastischen Konstanten einer plattenförmigen Struktur zu bestimmen.
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.
An approach to adhesive bond characterisation using guided acoustic waves in multi-layered plates
(2021)
An approach for the non-destructive characterisation of adhesive bonds using guided ultrasonic waves is presented. Pulsed laser radiation is used to thermoacoustically excite broadband ultrasonic waves in a multi-layered sample, consisting of a metal plate adhesively joined to a polymeric layer using synthetic resin. The resulting signals are received by a purpose-built piezoelectric transducer. Varying the distance between excitation and detection yields spatio-temporal measurement data, from which the dispersive properties of the propagating waves can be inferred using a two-dimensional Fourier transform, assuming the plates to act as coupled waveguides. Coupled multi-layered waveguides show an effect referred to as mode repulsion, where the distance between certain modes in the frequency-wavenumber domain is assumed to be a measure of coupling strength. Measurements at different stages of curing of the adhesive layer are performed and evaluated. A comparison of the results shows changes in the dispersive properties, namely an increased modal bandwidth for the fully cured sample as well as an increased modal distance.
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.
This presentation shows how the dispersive behavior of ultrasonic guided waves in isotropic materials can be used by means of Convolutional Neural Networks to determine the elastic parameters. For this purpose, the preprocessing, the training, the chosen architecture and the results are evaluated on the basis synthetic image data. This presentation was given at the SMSI 2021.
The use of composite materials has been steadily growing during the last decades, as well as the requirements on quality, mechanical properties and geometries of the parts. Some processes, like the 3D UV pultrusion process, manufacture parts that are long and thin, whilst having a varying curvature radius along the same part or even no curvature at all. Studying their mechanical properties along the main fiber direction, which is of foremost interest, is not an easy task nor efficient with most nondestructive methods. The use of air-coupled ultrasonics to evaluate the properties of composite materials has been widely proved by several authors, mainly using guided waves that provide information on the orthotropic properties of this kind of materials. Most of this work has focused on analyzing straight plate-like geometries, due to the simplicity to generate desired Lamb modes in the plate and analyze the behavior of guided waves inside the plate. In our contribution, the differences in the propagation of Lamb waves for straight and curved geometry glass fiber reinforced polymers (GFRP) have been analyzed. A GFRP test sample cured with UV light with one straight and one curved area has been evaluated. The responses of the generated Lamb wave modes for the straight and curved geometries have been compared, accounting for variations in the transducer characteristics, e.g. resonance behavior and focusing.
Practical applications of recent BAM-research results of NDT and monitoring in civil engineering
(2018)
Nondestructive testing and monitoring was of mayor interest in BAM-research related to the survey of civil engineering structures during the last two decades. Three case studies about research projects carried out at BAM were presented. All three projects focussed on the survey of parameters and safety related tasks in bridge engineering and inspection. First, the monitoring of the Berlin main station was presented with focus on the sensors developed specifically for this project. The second case study presented and application of guided ultrasonic waves to CFRP-strengthened bridges. The method was sucsessully applied to two of four different strengthening measures carried out min. 10 years ago. The last project presented the feasibility study of a network consisting of simple Bluetooth modules to survey changes in the moisture content in sand and massive concrete.