【摘要】 针对机械传动系统中采集的信号存在噪声干扰问题，提出了一种混合人工旅鼠算法(hybrid artificial lemming algorithm, HALA)优化变分模态分解(variational mode decomposition, VMD)与改进小波阈值去噪(improved wavelet threshold denoising, IWTD)的振动信号联合去噪方法。首先，通过HALA自适应选取VMD的关键参数，将含噪信号自适应分解为n个本征模态函数；其次，通过相关系数法筛选有效模态分量；最后，利用改进的小波阈值函数对选定分量进行二次去噪。结果表明：与VMD、小波阈值去噪(wavelet threshold denoising, WTD)、VMD-IWTD等去噪方法进行对比，基于HALA-VMD-IWTD的振动信号联合去噪方法去噪后的信号信噪比最高、均方根误差最小，具有更好的去噪优越性，适用于非平稳振动信号去噪；当故障特征频率为103.4 Hz时，经该方法去噪处理后，信号中的故障特征频率成分更加突出，背景噪声得到有效抑制。更多还原

【Abstract】 Here, aiming at the noise interference problem of signals collected in mechanical transmission systems, a vibration signals joint denoising method using hybrid artificial lemming algorithm(HALA) optimized variational mode decomposition(VMD) combined with improved wavelet threshold denoising(IWTD) was proposed. Firstly, key parameters of VMD were adaptively chosen with HALA to adaptively decompose signals with noise interference into n intrinsic mode function(IMF). Secondly, effective IMF components were screened out with the correlation coefficient method. Finally, the selected components were again denoised by using improved wavelet threshold function. Results showed that signals denoised with the proposed joint denoising method based on HALA-VMD-IWTD have the highest signal-to-noise ratio(SNR) and the lowest root mean square error(RMSE) compared to denoising methods based on VMD, wavelet threshold denoising(WTD) and VMD-IWTD; the proposed method has better denoising performance, particularly, for non-stationary vibration signals; when the fault feature frequency is 103.4 Hz, after using the proposed method, the component with the fault feature frequency is more prominent, background noise is effectively suppressed.更多还原