基于高斯白噪声加入的EMD在旋转机械故障诊断中的应用研究

Applied Research on Rotated Machine of Fault Diagnosis Based on EMD Added Gauss White Noise

为了避免经验模式分解（EMD）过程中不同时间尺度函数间的模式混叠，采用基于高斯白噪声加入的经验模式分解方法，并将之应用于旋转机械故障诊断中。该方法主要是对同一信号重复若干次加入相互独立的高斯白噪声序列，并分别筛选固有模式函数，而后把各个对应的固有模式函数进行平均计算，消除白噪声对分解结果的影响。最后，通过对固有模式函数进行包络解调，从中提取故障特征。对实际旋转机械故障振动信号的分析结果表明，该方法能有效避免固有模式函数间模式混叠，提高故障诊断效果。

The EMD added Gauss white noise is proposed to avoid mode mixing of different time-scale IMF, and is applied in fault diagnosis for rotating machine. The different time series of mutual-independent Gauss white noise is added into original s-ignal and repeated many times. Intrinsic mode function is sifted from signal added with white noise respectively. All of the corresponding IMF is averaged to eliminate influence on reresult brought by white noise. Ultimately, the faulty characteris-tic is achieved by demodulating the IMF. This method can eff-ectively eliminate mode mixing of IMF and improve the quality of fault diagnosis through analyzing vibrating signal of rotating machine.