Classification and prediction of multidamages in smart composite laminates using discriminant analysis

Abstract

A supervised machine learning framework is proposed for local assessments of delamination and transducer debonding in smart composite laminates while using their low-frequency structural vibrations. Load independent discriminative features were identified through a system identification algorithm and several supervised machine learning algorithms were employed to distinguish between the healthy and damaged structures. Linear discriminant analysis was shown to outperform other classifiers. The issue of overfitting of the training data was addressed by evaluating the predictive performance of the classifier on independent test cases. The proposed approach could help provide insightful guidelines for the assessment of multidamages in smart composite laminates.

Keywords:

• Delamination
• linear discriminant analysis
• sensor partial debonding
• supervised learning
• system identification

Availability of data

The data that support the findings of this study is available from the corresponding author upon request.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This research was supported by the Basic Science Research Program, through the National Research Foundation of Korea (NRF-2017R1D1A1B03028368 and 2020R1A2C1006613), funded by the Ministry of Education.