基于相关系数的EEMD转子振动信号降噪方法

Ensemble Empirical Mode Decomposition De-noising Method Based on Correlation Coefficients for Vibration Signal of Rotor System

针对转子振动信号周期性强的特点,应用集合经验模式分解(ensemble empirical mode decomposition,简称EEMD)对转子振动信号降噪过程中固有模式函数(intrinsic mode functions,简称IMF)分量的选取问题,提出了基于相关系数的EEMD降噪方法。首先,对原始信号进行EEMD分解得到IMF分量,并计算各IMF分量自相关函数与原信号自相关函数的相关系数;然后,根据相关系数选择相应的IMF分量重构信号最终达到对原信号降噪的目的;最后,对比了EEMD过程中不同加噪次数对降噪效率和效果的影响,给出了加噪次数的设置方法。仿真信号和转子振动信号的降噪结果表明了该降噪方法的可行性和有效性。

According to the problems of choosing the intrinsic mode functions(IMF) components when de-noising for vibration signal of rotor system by ensemble empirical mode decomposition(EEMD), and the vibration signal is periodic, an ensemble empirical mode decomposition de-noising method based on correlation coefficients is proposed. First of all, the ensemble empirical mode decomposition decomposes an original signal into a collection of intrinsic mode functions (IMFs), and calculates the correlation coefficients between IMFs’ autocorrelation function and original signal autocorrelation function. Second, IMF components are chosen to reconstruct signal based on correlation coefficients, and the original signal is de-noised. Meanwhile, the efficient and result are studied for different parameters of EEMD,then the method of choosing parameters is proposed. The simulate signal and vibration signal of rotor system are used to test and the results show that the de-noising method presented here has feasibility and validity.