Recognizing Running State of High-Speed Trains Based on Multifractal Theory and SVM

In order to evaluate in-service performances of high-speed trains, a novel approach to recognize the running state of high-speed trains was proposed using the multifractal theory and the support vector machine (SVM). The relationship between the multifractal spectrum parameters and the train running states was analyzed after the multifractal spectrum of the vibration signal was calculated by multifractal theory. Then, high-speed train running states were identified by SVM, using the characteristics of the multifractal spectrum width, the fractal dimension difference, and the spectrum skewness. In addition, a recognition experiment was carried out for three typical conditions of a certain type train, including the normal condition, the anti-hunting damper malfunction, and the air spring damper malfunction, after the SVM with different velocities and the SVM with a velocity (160 km/h) were trained using their multifratal characteristics. As a result, a state recognition accuracy of more than 88.8% was obtained, which verified the effectiveness of the proposed method.