Modern aerospace structures demand lightweight design procedures and require scheduled maintenance intervals. Supervised deep learning strategies can allow reliable damage detection provided a large amount of data is available to train. These learning algorithms may face problems in the absence of possible damage scenarios in the training dataset. This class imbalance problem in supervised deep learning may curtail the learning process and can possess issues related to generalization on unseen examples. On the other hand, unsupervised deep learning algorithms like autoencoders can handle such situations in the absence of labeled data. In this study, an aerospace composite panel is interrogated with a circular array of piezoelectric transducers using ultrasonic guided waves in a round-robin fashion. The time-series signals are collected for both the healthy and unhealthy state of the structure and transformed into a time-frequency dataset using continuous wavelet transformation. A convolutional autoencoder algorithm trained on healthy signals is used to identify anomalies in the form of delamination in the structure. The proposed methodology can successfully identify delamination in the structure with good accuracy.
The effect of temperature on guided waves is considered one of the crucial aspects of a structural health monitoring procedure. The influence of temperature can cause abrupt variations in the actual signatures and interfere with the existing damage identification strategies. In this paper, we have addressed this issue with a self-supervised deep learning-based temperature compensation methodology. We have used temperature affected time-traces from the dataset and converted them into 2D-representation using continuous wavelet transformation. We have proposed a new philosophy of temperature compensation in which the effect of temperature on the reference signals modifies amplitude and phase of the signal, is considered noise. We have trained a convolutional denoising autoencoder to transform temperature affected signals at any temperature into signals at the reference temperature. The performance of the algorithms is evaluated against unseen examples. It is seen that the proposed methodology can successfully compensate for temperature effects with a low mean squared loss and mean squared error and a high coefficient of determination.
KEYWORDS: Composites, Wave propagation, Finite element methods, Neural networks, Sensors, Structural health monitoring, Signal generators, Signal analysis, Sensor networks, Safety
Structural Health Monitoring (SHM) deals mainly with structures instrumented by secondary bonded or embedded sensors that, acting as both signal generators and receivers, are able to “interrogate” the structure about its “health status”. Sensorised structures appear promising for reducing the maintenance costs and the weight of aerospace composite structures, without any reduction of the safety level required. Much effort has been spent during last years on signal analysis techniques in order to extract from signals provided by the sensors networks many parameters, metrics, and images correlated to damages existence, location and extensions. As in many other technological fields, like medical image diagnostics, deep learning techniques in general and artificial neural networks in particular can be a very powerful instrument for damage patterns reconstruction and selection provided that a sufficient and consistent amount of data related to healthy and damaged configuration of the item under test are available. Within this work explicit finite element analysis has been employed to simulate waves propagation within composite plates with and without delaminations due to impacts. The numerical results have been previously validated with analytical solutions and experimental signals then have been used to populate the data sets necessary for deep learning. This paper will present the preliminary results achieved by the authors.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.