【摘要】 振动信号在噪声影响下,特征提取十分困难。为此应用同步挤压小波变换(Synchrosqueezing Wavelet Transform,SST)对振动信号进行降噪,针对分解后本征模态分量(Intrinsic Mode Function,IMF)的选取问题,提出一种基于瞬时频率复杂度和自相关系数峰度值的同步挤压小波变换降噪方法。算法首先对原始信号进行SST信号分解并提取小波脊线生成固有模态分量,然后对生成的分量进行Hilbert变换得到瞬时频率曲线,再根据瞬时频率的复杂度选择相应的合成分量重构信号。为了进一步消除噪声影响,该方法同时采用了自相关系数峰度阈值法对筛选后的分量进行二次剔除,最终实现对原始信号降噪的目的。试验最后通过不同标准方差的噪声仿真信号以及物流机械传送设备振动信号验证该方法的可行性和有效性,同时将该方法与基于集成经验模式分解(Ensemble Empirical Mode Decomposition,EEMD)和小波变换的方法进行比较,结果表明该方法的降噪性能要优于其他方法。更多还原

【Abstract】 The feature extraction from noise contaminated vibration signals is very difficult. The synchrosqueezed wavelet transform( SST) was introduced in the process of vibration signal denoising. To solve the difficulty of slecting intrinsic mode functions( IMF) s,a novel denoising method for vibration signals based on the instantaneous frequency complexity and autocorrelation coefficient kurtosis threshold of IMF was presented. The SST signal analysis method was ultilized to extract IMF components and then the Hilbert transform was applied to obtain the instantaneous frequency curve. IMF components were selected for reconstruction according to their corresponding component instantaneous frequency complexity. In order to further eliminate the noise effect,the correlation coefficient threshold method based on kurtosis value was employed to select once more the components for reconstructiing the original signal which can finally realize the original signal noise reduction. By experiments,the feasibility and effectiveness of the proposed method were verified by using simulated noise signals with different standard variances and bearing vibration signals of logistics machinery transmission equipments. Compared with other methods based on the ensemble empirical mode decomposition and wavelet transform,the results demonstrate that the proposed method is better than others.更多还原