Improving the Diagnostic Reliability of AE-based Failure Mode Detection and Distinction in CFRP

The usage of carbon fiber reinforced plastics (CFRP) in safety critical systems requires the application of Structural Health Monitoring (SHM). A well-known non-destructive testing (NDT) method is Acoustic Emission (AE). AE-based methods enable a continuous and in-situ monitoring of CFRP structures. While the classification of damages using AE signal features is thoroughly studied, the investigation of the correlation between the reliability of classification results and the applied loading patterns enables a new metric to ensure the success of SHM methods. In this contribution the probability of detection is used to evaluate the reliability of classification results. The four damage modes (debonding, delamination, matrix crack, and fiber breakage) are classified by a support vector machine (SVM). To distinguish the damage modes time-frequency domain features of the corresponding AE signal are calculated and finally classified to evaluate the dependency between the applied loading patterns and the classification quality. A concept of an online control loop is proposed using the reliability of classification results as control variable for improved testing strategies finally leading to ensure a safe usage of CFRP structures.

Keywords: Reliability, Diagnostics, Structural health monitoring, Acoustic emission, Online testing.

Download PDF