Filtern
Dokumenttyp
- Beitrag zu einem Tagungsband (26) (entfernen)
Schlagworte
- Lamb waves (6)
- Structural Health Monitoring (6)
- Ultrasonic Guided Waves (5)
- Composites (4)
- Materialcharakterisierung (3)
- Non-Destructive Testing (3)
- Ultrasonic guided waves (3)
- Ultrasound (3)
- Composite structures (2)
- Damage detection (2)
Organisationseinheit der BAM
Temperature Compensation Strategies for Lamb Wave Inspection using Distributed Sensor Networks
(2022)
The application of temperature compensation strategies is crucial in structural health monitoring approaches based on guided waves. Actually, 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 describes a method to compensate the temperature effect, namely the optimal baseline selection (OBS), extending its application to the case of distributed sensor networks (DSN). The effect of temperature separation between baseline time-traces in OBS are investigated considering multiple couples of sensors employed in the DSN. A combined strategy that uses both OBS and frequent value warning is considered. Theoretical results are compared, using 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 combination of results are available. Nonetheless, introducing a frequent value warning it is possible to increase the efficiency of the OBS approach making use of fewer signal processing algorithms. These confirm that the performance of OBS quantitatively agrees with predictions and also demonstrate that the use of compensation strategies improve detectability of damage.
Performance assessment for GuidedWave (GW)-based Structural Health Monitoring (SHM) systems is of major importance for industrial deployment.
With conventional feature extraction methods like damage indices, pathbased probability of detection (POD) analysis can be realized. To achieve reliability quantification enough data needs to be available, which is rarely the case.
Alternatives like methods for performance assessment on system level are still in development and in a discussion phase. In this contribution, POD results using an Artificial Intelligence (AI)-based data analysis are compared with those delivered by conventional data analysis. Using an open-access dataset from Open Guided Wave platform, the possibility of performance assessment for GW-based SHM systems using AI-based data analysis is shown in detail. An artificial neural network (ANN) classifier is trained to detect artificial damage in a stiffened CFRP plate. As input for the ANN, classical damage indicators are used. The ANN is tested to detect damage at another position, whose inspection data were not previously used in training. The findings show very high detection capabilities without sorting any specific path but only having a global view of current damage metrics. The systematic evaluation of the ANN predictions with respect to specific damage sizes allows to compute a probability of correct identification versus flaw dimension, somehow equivalent to and compared with the results achieved through classic path-based POD analysis. Also, sensitive paths are detected by ANN predictions allowing for evaluation of maximal distances between path and damage position. Finally, it is shown that the prediction performance of the ANN can be improved significantly by combining different damage indicators as inputs.
Reliability assessment of Structural Health Monitoring (SHM) systems
poses new challenges pushing the research community to address many questions which are still open. For guided wave-based SHM it is not possible to evaluate the system performance without taking into account the target structure and applied system parameters. This range of variables would result in countless measurements.
Factors like environmental conditions, structural dependencies and wave characteristics demand novel solutions for performance analysis of SHM systems compared to those relying on classical non-destructive evaluation. Such novel approaches typically require model-assisted investigations which may not only help to explain and understand performance assessment results but also enable complete studies without costly experiments. Within this contribution, a multi input multi output approach using a sparse transducer array permanently installed on a composite structure to excite and sense guided waves is considered. Firstly, the method and the analysis of path-based performance assessment are presented considering an open-access dataset from the Open Guided Wave platform. Then, a performance analysis of a guided wave-based SHM system using Probability of Detection is presented. To explain some unexpected results, the model-assisted investigations are used to understand the physical phenomena of wave propagation in the test specimen including the interaction with damage. Finally, issues and future steps in SHM systems’ performance assessment and their development are discussed.
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.
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.
Бегущие упругие волны часто используются в области неразрушающего контроля для определения механических характеристик материалов. Поскольку склеивание различных материалов является широко используемым методом в автомобильной и авиационной промышленности, необходима надежная система измерения для определения качества таких клеевых соединений. Дисперсионные кривые для многослойных волноводов при наличии клеевых прослоек характеризуются появлением эффекта расталкивания нормальных мод в тех областях, где соответствующие моды для отдельных материалов пересекались бы. Таким образом, в зависимости от качества клеевого соединения расстояние между этими модами изменяется. Используя комбинацию экспериментальных и численных данных, в настоящей работе представлен подход к определению параметра, указывающего на качество клеевого соединения.
Detailliertes Wissen über die mechanischen Eigenschaften verwendeter Materialien ist Grundvoraussetzung für viele ingenieurtechnische Aufgaben und Dienstleistungen. Zur Bestimmung der elastischen Materialparameter gibt es verschiedene klassische, zerstörende Prüfverfahren. Eine Möglichkeit der zerstörungsfreien Bestimmung liegt in der Auswertung von Messergebnissen, die auf Basis des Ausbreitungsverhaltens geführter Ultraschallwellen gewonnen wurden. Das Ausbreitungsverhalten geführter Ultraschallwellen kann mittels Dispersionsabbildungen dargestellt werden.
Um aus messtechnisch ermittelten Dispersionsabbildungen 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 Beitrag wird gezeigt, wie das Ausbreitungsverhalten von geführten Ultraschallwellen unter Verwendung von CNNs genutzt werden kann, um die isotropen elastischen Konstanten einer plattenförmigen Struktur zu bestimmen. Hierfür werden die verwendeten Daten analysiert, das Preprocessing erläutert und eine grundlegende CNN-Architektur gewählt. Zur Auswertung des generierten Modells werden verschiedene Verfahren wie Gradienten-Mapping und die Visualisierung der verschiedenen Schichten vorgestellt. Die Anwendbarkeit der Methode wird anhand synthetischer Daten demonstriert.
In the field of non-destructive testing, Lamb waves are often used for material characterisation. The increasing computational capabilities further enable complex and detailed simulations to predict the material behaviour in reality. Since adhesive bonding of different materials is a widely used method in modern applications, a reliable measurement system to determine the quality of these adhesive bonds is needed. Investigations of the dispersive behaviour of acoustic waves in adhesively bonded multi-layered waveguides show mode repulsions in the dispersion diagrams in regions where the modes of the single materials would otherwise intersect. In previous works, changes of the distance between those modes with respect to the bonding quality are observed. The experimental data for this investigation is generated using pulsed laser radiation to excite broadband acoustic waves in plate like specimens which are detected by a piezoelectric ultrasonic transducer. Numerical data is generated using simulations via a semi-analytical finite element method. Using a combination of experimental and numerical data, the present work introduces an approach to determine a parameter which indicates the bonding quality in relation
to an ideal material coupling.
In the context of Industry 4.0 and especially in the field of Structural Health Monitoring, Condition Monitoring and Digital Twins, simulations are becoming more and more important. The exact determination of material parameters is required for realistic results of numerical simulations of the static and dynamic behavior of technical structures. There are many possibilities to determine
elastic material parameters. One possibility of non-destructive testing are ultrasonic guided waves. For the evaluation of the measurement results, mostly inverse methods are applied in order to be able to draw conclusions about the elastic material parameters from analysing the ultrasonic guided wave propagation. For the inverse determination of the elastic material Parameters with ultrasonic guided waves, several investigations were carried out, e.g. the determination of the isotropic material parameters through the point of zero-groupvelocity or anisotropic material parameters with a simplex algorithm. These investigations are based on the evaluation of dispersion images. Machine learning and in particular Convolutional Neural Networks (CNN) are one possibility of the automated evaluation from Image data, e.g. classification or object recognition problems. This article shows how the dispersive behavior of ultrasonic guided waves and CNNs can be used to determine the isotropic elastic constants of plate-like structures.
Lamb waves are a common tool in the field of non-destructive testing and are widely used for materialcharacterisation. Further, the increasing computational capability of modern systems enables the Simulation of complex and detailed material models. This work demonstrates the possibility of simulating an adhesive-bonded multilayer system by characterising each layer individually, and introduces an Approach for determining the dispersive behaviour of acoustic waves in a multilayer system via real measurements.