Hidden Defects Diagnosis Using Parameter Optimization

In this paper, the issue of composite defects diagnosis by applying the support vector machine (SVM) was addressed. The component analysis was performed initially to extract the features and to reduce the dimensionality of original data features. Kernel parameters selection of support vector machine which has great influence on the performance of defects classification has been discussed in this work. Precisely, we focus on wavelet transform to extract the feature from the original signals, adopt component analysis to do feature selection and apply support vector machine to classify the defects. This paper exploits the parameter optimization procedure to ensure the generalization ability of SVM. The result shows that multi-class SVM produces promising results and has the potential for use in fault diagnosis.