---

概要

This study develops a computational algorithm to distinguish stress corrosion cracking and fatigue cracks from signals measured electromagne tic-nondestructively. The algorithm utilizes k-nearest neighbour method, and distinguishes the two types of crack on the basis of five-dimens ional features vector extracted from the signals. Experimental verification are performed using 82 specimens including 45 artificial stress corrosion cracking and 37 artificial fatigue cracks. The results demonstarts that the algorithm can correctly classify almost all the cracks.

---

全文掲載のPDFファイルダウンロード

• PDFファイルダウンロード