Abstract: A new method of detecting dynamic imbalance with cardan shaft in the high-speed train is proposed applying the combination between ensemble empirical model decomposition(EEMD), Hankel matrix and singular value decomposition(SVD) contrary to the aliasing defect between the adjacent intrinsic model functions existing in the EEMD. The vibration acceleration signals of gimbal are decomposed through EEMD to get the different intrinsic model components. The Hankel matrix, which is constructed throng the single decomposition model component, is orthogonally executed through SVD. The key singular values are selected to reconstruct vibration signs on the base of the key stack of singular values. Fourier spectrum of the reconstructed signal is applied to detect dynamic imbalance with shaft and eliminates clutter spectrum caused by the aliasing defect between the adjacent intrinsic model functions, highlights the failure characteristics. The method is verified by test data in the condition of dynamic imbalance, the results show this method can effectively detect the fault vibration characteristics caused by cardan shaft dynamic imbalance and extract the nature vibration features. With comparison to the simple EEMD, the clarity and failure characterization force are significantly improved.

Key words: dynamic detection, dynamic imbalance with cardan shaft, ensemble empirical model decomposition(EEMD), Hankel matrix, high-speed train, singular value decomposition(SVD)