Damage identification using wave damage interaction coefficients predicted by deep neural networks

Humer, Christoph; Hoell, Simon; Kralovec, Christoph; Schagerl, Martin

doi:10.1016/j.ultras.2022.106743

Cited by 21 publications

(6 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To reduce this large computational effort, the application of numerical surrogate models (e.g., ML models) can be used to lower the number of required simulations by interpolation. An approach was recently demonstrated for the accurate interpolation of simulated guided wave signals by Humer [ 33 ]. The available simulated signals (and their correlated and known damage parameters) enable very fast damage identification, by simply comparing their similarity to measured EMI spectra.…”

Section: Introductionmentioning

confidence: 99%

Sandwich Face Layer Debonding Detection and Size Estimation by Machine-Learning-Based Evaluation of Electromechanical Impedance Measurements

Kralovec

Lehner

Kirchmayr

et al. 2023

Sensors

Self Cite

View full text Add to dashboard Cite

The present research proposes a two-step physics- and machine-learning(ML)-based electromechanical impedance (EMI) measurement data evaluation approach for sandwich face layer debonding detection and size estimation in structural health monitoring (SHM) applications. As a case example, a circular aluminum sandwich panel with idealized face layer debonding was used. Both the sensor and debonding were located at the center of the sandwich. Synthetic EMI spectra were generated by a finite-element(FE)-based parameter study, and were used for feature engineering and ML model training and development. Calibration of the real-world EMI measurement data was shown to overcome the FE model simplifications, enabling their evaluation by the found synthetic data-based features and models. The data preprocessing and ML models were validated by unseen real-world EMI measurement data collected in a laboratory environment. The best detection and size estimation performances were found for a One-Class Support Vector Machine and a K-Nearest Neighbor model, respectively, which clearly showed reliable identification of relevant debonding sizes. Furthermore, the approach was shown to be robust against unknown artificial disturbances, and outperformed a previous method for debonding size estimation. The data and code used in this study are provided in their entirety, to enhance comprehensibility, and to encourage future research.

show abstract

Section: Introductionmentioning

confidence: 99%

Sandwich Face Layer Debonding Detection and Size Estimation by Machine-Learning-Based Evaluation of Electromechanical Impedance Measurements

Kralovec

Lehner

Kirchmayr

et al. 2023

Sensors

Self Cite

View full text Add to dashboard Cite

show abstract

“…At the cost of significant computational power requirements, numerical methods allow for complex real case scenario simulation. Recent advances covers in particular: Guided waves scattering on impact damage of composite structures [ 43 , 47 ] or delaminations [ 48 ]; Transmission of guided waves across partially-closed cracks [ 49 ]; Wave damage interaction coefficients for lightweight structures [ 10 ]; Damage of reinforced concrete beams [ 50 ]; Looseness of joint structures in cylindrical waveguides [ 51 ]; Matrix cracking in laminated composites [ 52 ]. …”

Section: Materials and Methodsmentioning

confidence: 99%

“…Unfortunately, analyzing these output signals is a non-trivial task due to the complexity of the output. This naturally motivates research in applying the latest advances in machine learning for processing and interpreting these signals [ 9 , 10 , 11 , 12 , 13 , 14 ]. However, machine learning generally requires vast amounts of data to be appropriately trained.…”

Section: Introductionmentioning

confidence: 99%

Diagnostic-Quality Guided Wave Signals Synthesized Using Generative Adversarial Neural Networks

Heesch

Dziendzikowski

Mendrok

et al. 2022

Sensors

View full text Add to dashboard Cite

Guided waves are a potent tool in structural health monitoring, with promising machine learning algorithm applications due to the complexity of their signals. However, these algorithms usually require copious amounts of data to be trained. Collecting the correct amount and distribution of data is costly and time-consuming, and sometimes even borderline impossible due to the necessity of introducing damage to vital machinery to collect signals for various damaged scenarios. This data scarcity problem is not unique to guided waves or structural health monitoring, and has been partly addressed in the field of computer vision using generative adversarial neural networks. These networks generate synthetic data samples based on the distribution of the data they were trained on. Though there are multiple researched methods for simulating guided wave signals, the problem is not yet solved. This work presents a generative adversarial network architecture for guided waves generation and showcases its capabilities when working with a series of pitch-catch experiments from the OpenGuidedWaves database. The network correctly generates random signals and can accurately reconstruct signals it has not seen during training. The potential of synthetic data to be used for training other algorithms was confirmed in a simple damage detection scenario, with the classifiers trained exclusively on synthetic data and evaluated on real signals. As a side effect of the signal reconstruction process, the network can also compress the signals by 98.44% while retaining the damage index information they carry.

show abstract

“…Liu and Zhang [19] converted the Lamb wave signal into an image, and the deep CNN was used to process the converted image, the notch-type crack damage detection in thin metal plates was achieved, and they experimentally verified that the method had a significant improvement in detection and classification accuracy compared with the traditional neural network. The deep neural network (DNN) was used by Humer et al [20] to enhance the wave damage interaction coefficients database for realizing the prediction of plate structure damage identification. Rai and Mitra [21] developed a deep learning model with multi-head onedimensional convolutional neural network, which can directly operate the obtained discrete time domain Lamb wave signals and accurately predict the damage.…”

Section: Introductionmentioning

confidence: 99%

SN-SAE: a new damage diagnosis method for CFRP using Lamb wave

Wang

et al. 2023

Meas. Sci. Technol.

View full text Add to dashboard Cite

The damage diagnosis of carbon fiber reinforced polymer (CFRP) using Lamb wave has been widely developed, but it is still a challenging task to obtain reliable damage diagnosis results by analysis of Lamb wave, the emergence of deep learning models provides an effective solution for this work. However, the internal covariate shift and overfitting exist in traditional deep networks. The SN-SAE deep neural network model is proposed by introducing stochastic normalization (SN) into stacked autoencoder (SAE). The signals of 28 different damage locations in the CFRP plate provided by the open platform are processed by SN-SAE, and the damage diagnosis at different locations is achieved. The validity of SN-SAE was further verified by data obtained through building an experimental platform. The results demonstrated that the SN-SAE model can achieve high test accuracy with only 15% of the data samples as training with limited data sample, which provides a simple and effective solution for damage diagnosis of composite plates.

show abstract

Damage identification using wave damage interaction coefficients predicted by deep neural networks

Cited by 21 publications

References 42 publications

Sandwich Face Layer Debonding Detection and Size Estimation by Machine-Learning-Based Evaluation of Electromechanical Impedance Measurements

Sandwich Face Layer Debonding Detection and Size Estimation by Machine-Learning-Based Evaluation of Electromechanical Impedance Measurements

Diagnostic-Quality Guided Wave Signals Synthesized Using Generative Adversarial Neural Networks

SN-SAE: a new damage diagnosis method for CFRP using Lamb wave

Contact Info

Product

Resources

About