22 April 2016 Lamb wave feature extraction using discrete wavelet transformation and Principal Component Analysis
Author Affiliations +
In this research, a new method is presented for eliciting the proper features for recognizing and classifying the kinds of the defects by guided ultrasonic waves. After applying suitable preprocessing, the suggested method extracts the base frequency band from the received signals by discrete wavelet transform and discrete Fourier transform. This frequency band can be used as a distinctive feature of ultrasonic signals in different defects. Principal Component Analysis with improving this feature and decreasing extra data managed to improve classification. In this study, ultrasonic test with A0 mode lamb wave is used and is appropriated to reduce the difficulties around the problem. The defects under analysis included corrosion, crack and local thickness reduction. The last defect is caused by electro discharge machining (EDM). The results of the classification by optimized Neural Network depicts that the presented method can differentiate different defects with 95% precision and thus, it is a strong and efficient method. Moreover, comparing the elicited features for corrosion and local thickness reduction and also the results of the two’s classification clarifies that modeling the corrosion procedure by local thickness reduction which was previously common, is not an appropriate method and the signals received from the two defects are different from each other.
© (2016) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Mojtaba Ghodsi, Mojtaba Ghodsi, Hamidreza Ziaiefar, Hamidreza Ziaiefar, Milad Amiryan, Milad Amiryan, Farhang Honarvar, Farhang Honarvar, Yousef Hojjat, Yousef Hojjat, Mehdi Mahmoudi, Mehdi Mahmoudi, Amur Al-Yahmadi, Amur Al-Yahmadi, Issam Bahadur, Issam Bahadur, "Lamb wave feature extraction using discrete wavelet transformation and Principal Component Analysis", Proc. SPIE 9804, Nondestructive Characterization and Monitoring of Advanced Materials, Aerospace, and Civil Infrastructure 2016, 98041F (22 April 2016); doi: 10.1117/12.2218842; https://doi.org/10.1117/12.2218842

Back to Top