• CN: 11-2187/TH
  • ISSN: 0577-6686

Journal of Mechanical Engineering ›› 2026, Vol. 62 ›› Issue (16): 29-40.doi: 10.3901/JME.260733

Previous Articles    

Quantitative Research on Shimmy of EMU Car Body Driven by Physical Model and Data Fusion

DENG Xin1,2, GONG Dao1,2, ZHOU Kai1,2, ZHOU Jinsong1,2, WANG Ruiqian3, SUN Weiguang1,2,4   

  1. 1. College of Transportation, Tongji University, Shanghai 201800;
    2. Shanghai Key Laboratory of Rail Infrastructure Durability and System Safety, Shanghai 201800;
    3. School of Mechanical and Rail Transit Engineering, Changzhou University, Changzhou 213164;
    4. CRRC Qingdao Sifang Co., Ltd., Qingdao 266111
  • Received:2025-09-12 Revised:2026-03-06 Published:2026-10-14

Abstract: The carbody chattering of EMUs is caused by the coupling of bogie hunting and carbody diamond-shape mode, which seriously affects running stability. Real-time monitoring of the chattering status is crucial for driving safety. Given that traditional methods struggle to accurately distinguish the"non-chattering, incubation, and chattering" stages, a chattering quantification method integrating physical models and data is proposed. At the physical level, through multi-source test analysis, it is revealed that the reduced damping ratio of bogie in-phase hunting and the decreased modal frequency difference induce coupled vibration, thereby triggering chattering. At the data level, improved variational mode decomposition(VMD) and Hilbert transform(HT) are adopted to extract instantaneous vibration features. Four-dimensional core indicators are constructed to characterize the chattering evolution chain,including modal energy-gathering fingerprintRE, modal damping fingerprint Db, modal interlocking coefficient Lm, and modal coherence coefficient Pfr. By fusing 500 sets of data (200 sets of simulation data, 200 sets of measured data, and 100 sets of augmented data), kernel density estimation(KDE) is used to model the feature distributions of the three stages. Combined with Bayesian decision-making and the “duration gate plus multi-evidence consistency gate”, false alarms are reduced. Experiments show that the proposed indicators can effectively distinguish the three stages of chattering, and the model's judgment is highly consistent with actual working conditions, providing a quantitative basis for online monitoring and suppression of carbody chattering.

Key words: train, modal, data fusion, modal characteristic index, index quantification

CLC Number: