基于EMD相关去噪的风电机组振动噪声抑制及特征频率提取

针对风电机组振动信号同时受背景白噪声和短时干扰噪声的影响,使得早期微弱故障特征频率难以提取的问题,提出一种结合经验模态分解(EMD)、相关性分析和小波包变换(WPT)的振动信号噪声抑制及故障特征频率提取方法(EMD相关去噪-WPT)。该方法首先利用EMD分解振动信号得到能表征不同频率的固有模态函数(IMF),然后筛选表征故障特征频率的IMF,并重构得到故障特征信号;其次,利用自相关分析去除重构信号中噪声的影响;最后,结合小波包变换(WPT)提取去噪重构振动信号中的特征频率。为了验证所提方法的有效性,以实测和模拟的双馈风电机组轴承故障振动信号为例,对轴承振动信号分别利用小波包变换(WPT)、EMD相关去噪-WPT、小波硬阀值-WPT方法进行特征频率提取分析。通过不同特征频率提取方法比较表明,所提出的基于EMD相关去噪-WPT特征频率提取方法,能够更有效地抑制背景白噪声和短时干扰噪声的影响,提取出早期微弱故障特征。

Noise suppression and characteristic frequency extraction of wind turbine vibration based on EMD correlation denoising

frequency extraction method combining empirical mode decomposition ( EMD) , correlation analysis with wavelet package transform ( WPT) were studied. This method, firstly, decomposes the vibration signals into a series of intrinsic mode functions ( IMFs) which represent different frequencies by using EMD. Then, a fault characteristic signal was restructured by accumulating the selected IMFs which characterize the fault characteristic frequencies. Secondly, the characteristic signals were analyzed by using the meth-od of autocorrelation analysis to eliminate the interference of the noises. Finally, the characteristic fre-quency is extracted by using the WPT from de-noising restructured vibration signals. WPT, EMD correla-tion denoising-WPT and wavelet hard thresholding-WPT were used to analyze the actual and simulating wind turbines bearing fault vibration signals to verify the effectiveness of the proposed method. The results of comparing with the different characteristic frequency extraction methods show that the presented charac-teristic frequency extraction method based on EMD correlation denoising-WPT can effectively depress the white noise and short-term disturbance noise, and extract early weak fault feature.