Cracking Acoustic Emission Identification Method based on PCA and SVM

Information redundancy is a big problem in acoustic emission(AE) signal identification based on statistical feature parameters such as amplitude, energy counts, etc. Here, principle component analysis(PCA) was employed to reduce information redundancy and extract statistical feature of AE signals.The AE data were collected in the AE test for Cr-coating cracking on the surface of a steel plate, AE statistical feature parameters were analyzed using PCA, and two principle components were extracted. The principle components were employed as the input vector of a SVM classifier, and the AE signals caused by Cr-coating cracking were identified . It demonstrated that principle components could represent statistical feature of AE signals, reduce information redundancy, and effectively raise identification efficiency and accuracy.