Small Amplitude Hunting Instability of High-speed Train Diagnosis Method Based on Modified Ensemble Empirical Mode Decomposition, Shannon Entropy and Least Square Support Vector Machine

Abstract

To monitor the state of small hunting instability for the train at a high speed, aiming at the problem of mode splitting of ensemble empirical mode decomposition (EEMD), a new methodology which combines modified ensemble empirical mode decomposition (MEEMD), Shannon entropy feature and least squares support vector machine (LSSVM) is presented in this paper to diagnose hunting motion state of high- speed train. Firstly, the vibration signal under 330Km/h~350Km/h is decomposed by MEEMD. Then, calculating the Shannon feature of IMFs and using LSSVM to recognize the hunting motion state. The result shows that the methodology of MEEMD-Shannon features-LSSVM can accurately recognize the unsteady state of hunting motion, the recognition rate is up to 97.78%. Furthermore, the accuracy and computation time are superior to ensemble empirical mode decomposition- support vector machine (EEMD-SVM). KEYWORD: high-speed train, small hunting, MEEMD, Shannon entropy, diagnose