Dynamic unbalance detection of Cardan shaft in high-speed train applying double decomposition and double reconstruction method. (September 2015)

• Record Type: Journal Article
• Authors: Ding, Jianming; Lin, Jianhui; Yu, Shengwei
• Is Part Of: Measurement. Volume 73(2015:Sep.)
• Page Start: 111
• Page End: 120
• Publication Date: 2015-09
• Subjects: High-speed train -- Cardan shaft -- Wavelet decomposition -- SVD -- DDDR -- Dynamic unbalance detection; Weights and measures -- Periodicals; Measurement -- Periodicals; Measurement; Weights and measures; Periodicals; 530.8
• DOI: 10.1016/j.measurement.2015.05.016
• Languages: English
• ISSNs: 0263-2241

Abstract

Highlights: DDDR is a newly combination model of wavelet decomposition and SVD. DDDR eliminates the frequency overlap influence between adjacent scales in wavelet domain. DDDR efficiently seizes weak fundamental, multiplier, divider frequency fault feature. Fourier spectrum clarity of DDDR processing is significantly improved. Abstract: The unbalance of Cardan shaft compromises operations of high-speed train. A new method is proposed to detect the unbalance by applying DDDR (double decomposition and double reconstruction method). The vibration acceleration of gearbox was decomposed into eight scale wavelet coefficients through wavelet packet decomposition. The eight single scale vibration signals were reconstructed by the corresponding scale wavelet coefficients. Hankel matrices in different scales were constructed through the reconstructed vibration signals in wavelet domain. SVD (singular value decomposition) of Hankel matrices was executed, and critical singular values were selected based on the maximum change of singular values. Those selected singular values were used to reconstruct the single scale vibration signal. So far, DDDR processing of signal has been completed. Fourier spectrum of signal acquired by DDDR processing was used to detect dynamic unbalance of high-speed train Cardan shaft. The validity of this method is supported by experimental data collected on dynamic unbalance experiments. The results show that this method can effectively extract the vibration characteristics of fundamental, multiplier, and divider frequencies. With comparison to the traditional wavelet decomposition, wavelet singularity value decomposition, the clarity and sensitive force have been significantly improved.