基于白噪声统计特性与EEMD的高速列车横向减振器故障诊断*

针对高速列车横向减振器故障信号非线性非平稳的特点，提出了基于白噪声统计特性与聚合经验模态分解（EEMD）相结合的故障诊断算法。首先，利用经验模态分解（EMD）对故障信号进行去噪，然后对去噪后的信号进行EEMD分解，最后对用相关系数求得的最能反映振动信号的本征模态函数（IMF）计算排列组合熵。在240km/h速度下，对高速列车横向减振器7种工况进行诊断，识别率达到91.8%。实验结果表明：与基于小波熵特征分析的算法相比，该算法具有更高的识别率和更强的抗噪性能。

Lateral damper fault diagnosis of high-speed train based on the statistical characteristics of white noise and EEMD

Considering the nonlinearity and nonstationarity of the lateral damper fault signal of high-speed train, a fault diagnosis method was proposed by combining ensemble empirical mode decomposition (EEMD) with the characteristics of white noise. Empirical mode decomposition (EMD) was used to denoise the original fault signal, and then the denoised signal was decomposed by EEMD. The permutation entropy of the IMF which best fits the original signal based on correlation analysis, was calculated. The recognition rate reaches 91.8% when applying this algorithm to diagnose seven different lateral damper faults of the high-speed train at the speed of 240km/h. This algorithm has higher recognition rate and stronger anti-noise performance comparing with wavelet entropy feature analysis method.