关键词: 高速列车, 小幅蛇行, S变换, 小波变换, 最小二乘支持向量机

Abstract: For the problems that the existing high-speed train monitoring methods did not consider the non-stationary characteristics of vibration signals of the high-speed trains, which caused the trains to lose the hunting stability, based on S-transform a method was proposed to process the high-speed train bogie signals and then extract the features. The LS-SVM was utilized to train and identify the features. The results show that the recognition accuracy of the features based on S-transform is 100%, which is much better than that of the wavelet-transform-based method. Therefore, the proposed method may predict the running states of the high-speed trains in time and ensure the operation security of the trains.

Key words: high-speed train, small hunting, S-transform, wavelet transform, least square support vector machine（LS-SVM）