基于GWO-NLM与CEEMDAN的滚动轴承故障诊断方法

Rolling bearing fault diagnosis method based on GWO-NLM and CEEMDAN

针对滚动轴承故障振动信号受背景噪声干扰大、故障特征不易提取的问题,提出了基于灰狼算法(GWO)优化的非局部均值去噪(NLM)和完全自适应噪声集合经验模态分解(CEEMDAN)相结合的轴承故障诊断方法。先将CEEMDAN和相关系数-能量比-峭度准则作为预处理手段,并进行信号重构;然后使用灰狼算法对NLM的参数进行优化,利用最优参数对重构信号进行降噪,将降噪后的信号通过SG(SavitzkyGolay)滤波进行二次降噪,得到最终去噪信号,对最终信号进行包络分析得到诊断结果。GWO-NLM去噪、CEEMDAN和包络分析的混合特征提取技术,由仿真信号可知去噪后的信噪比提高了9.31 dB,由实验信号可知能清晰地提取轴承的故障特征频率及倍频、转频以及故障特征频率与转频的系列调制频率。

For the problem that the vibration signal of rolling bearing faults is disturbed by background noise and the fault features are not easily extracted,a combination of non-local mean denoising(NLM)based on the optimization of the gray wolf algorithm(GWO)and fully adaptive noise-enabled ensemble empirical modal decomposition(CEEMDAN)was proposed for bearing fault diagnosis.First,CEEMDAN and the Correlation coefficient-energy ratio-kurtosis criterion were used as preprocessing way,and signal reconstruction was performed;then the grey wolf algorithm was used to optimize the parameters of NLM,and the optimal parameters were used to denoise the reconstructed signal,and secondary denoising of the denoised signal was achieved through SG(Savitzky-Golay)filtering to obtain the final denoised signal,and envelope analysis of the final signal was performed to obtain diagnostic results.For the hybrid feature extraction technique of GWO-NLM denoising,CEEMDAN and envelope analysis,the signal-to-noise ratio was improved by 9.31 dB after denoising as shown by the simulated signal,and the fault characteristic frequency and multiplication frequency of the bearing and the series modulation frequency of the fault characteristic frequency and rotation frequency can be clearly extracted by the experimental signal.