PSI - Issue 52

Yuhang Pan et al. / Procedia Structural Integrity 52 (2024) 699–708 Author name / Structural Integrity Procedia 00 (2019) 000 – 000

705

7

4. Results and discussion In this section, the proposed method is firstly used to detect the damage on the aluminum, and the change of the damage index is also showed in this section. Then, the BPNN is used to localize the damage based on different features. 4.1. Damage detection To evaluate the effectiveness of the damage index for damage detection, the change of damage index of guided wave method before and after damage is summarized in Fig.6. Taking the damage index obtained by Eq (1) as example, Fig.6(a) shows the damage index at pristine state, with values less than 0.01 on all paths. Fig.6(b) shows the damage index obtained from the vicinity of sensing paths 1-2 and 1-4, and it can be seen that the value of the damage index on the sensing paths close to the damage location increases a lot when the damage is introduced. This indicates the effectiveness of the damage index used in this study.

Fig. 6. Damage index 1 obtained different condition: (a) Pristine state, (b) Damage state.

A total of 160 sets of data were obtained from the experiment, including 144 sets of data from different damage locations and 16 sets of data from the pristine state. 30 sets of data (15 damage samples and 15 pristine samples) were selected for the validation of damage detection model, where 20 sets of data (10 damage samples, 10 pristine samples) were used for the training of the neural pattern recognition model, and 10 sets of data (5 damage samples and 5 pristine samples) were used as the testing of the neural network, and the results of the testing based on different damage index are summarized in Fig. 7, which shows all of three damage indices extracted from the guided wave method are capable of achieving 100% accuracy for the damage detection. This indicates the effectiveness of the method used in this research.

Fig. 7. Damage detection results obtained from: (a) Eq. (1), (b) Eq. (2), (c) Eq. (3).