Automatic wedge tightness classifying system by support vector machine

Abstract

This paper introduces a newly developed automatic classification system for wedge tightness inside the generator by applying support vector machine (SVM) classifier. The automatic classifying system for wedge tightness of the generator consists of 4 parts including data collection, preprocessing, feature extraction, and classification. Machine learning algorithm called SVM is used with the linear and radial basis function (RBF) classifier. Each input feature is extracted in different ways to evaluate the performance of classification. The evaluation is completed by using a 10-fold cross validation technique to provide high accuracy and a low number of False Negatives (FN). By applying the proposed system, the number of tightness and looseness inside wedge generator can be classified. Based on the classification results, the signals extracted in the frequency domain gives the best performance among the time domain and the frequency domain. This paper shows that the automatic classifying method has a high potential to identify the wedge tightness inside the generator.