ePrints@IIScePrints@IISc Home | About | Browse | Latest Additions | Advanced Search | Contact | Help

Deep Learning Based Identification of Elastic Properties Using Ultrasonic Guided Waves

Gopalakrishnan, K and Rautela, M and Deng, Y (2021) Deep Learning Based Identification of Elastic Properties Using Ultrasonic Guided Waves. In: 10th European Workshop on Structural Health Monitoring, EWSHM 2020, 1 Jul 2020, pp. 77-90.

Full text not available from this repository. (Request a copy)
Official URL: https://doi.org/10.1007/978-3-030-64908-1_8


Identification of elastic properties is relevant for both non-destructive material characterizations as well as for in-situ condition monitoring to predict any possible material degradation. In this paper, we have proposed deep learning frameworks to solve the inverse problem of material property identification using ultrasonic guided waves. The propagation of guided waves in a composite laminate is modelled using a reduced order Spectral Finite Element Method (SFEM). We have used two fundamental modes of a guided wave i.e. the anti-symmetric (A0) and the symmetric modes (S0) as inputs for the proposed deep learning models and elastic properties of a unidirectional composite laminate as outputs. The deep regression-based networks like, 1D-Convnets and LSTMs are used to map input space to target space. The performance of the algorithms is evaluated based on the mean squared loss value, coefficient of determination, and mean absolute error. It is seen that the networks can learn the inverse mapping and generalize well to unseen examples even in the presence of noise at various levels. This novel methodology can eliminate disadvantages associated with existing global optimization techniques in terms of accuracy, robustness, scale, and computational time. © 2021, The Author(s), under exclusive license to Springer Nature Switzerland AG.

Item Type: Conference Paper
Publication: Lecture Notes in Civil Engineering
Publisher: Springer Science and Business Media Deutschland GmbH
Additional Information: The copyright for this article belongs to Springer Science and Business Media Deutschland GmbH
Keywords: Condition monitoring; Convolutional neural networks; Elasticity; Global optimization; Guided electromagnetic wave propagation; Inverse problems; Laminated composites; Structural health monitoring; Ultrasonic testing; Ultrasonic waves, Coefficient of determination; Global optimization techniques; Learning based identification; Material characterizations; Property identification; Spectral finite element method; Ultrasonic guided wave; Unidirectional composites, Deep learning
Department/Centre: Division of Mechanical Sciences > Aerospace Engineering(Formerly Aeronautical Engineering)
Date Deposited: 23 Mar 2021 10:15
Last Modified: 23 Mar 2021 10:15
URI: http://eprints.iisc.ac.in/id/eprint/68549

Actions (login required)

View Item View Item