针对柴油机曲轴轴承声发射(Acoustic Emission,AE)信号中裂纹特征信息微弱,易与噪声混淆等问题,在K-SVD字典对信号稀疏的基础上,提出一种均值信号改进的K-SVD字典的滑动轴承AE信号去噪算法.重组均值信号和扩展到K-SVD信号矩阵中,在实现K-SVD稀疏AE信号矩阵的同时,稀疏提取均值信号的裂纹信号,利用K-SVD处理前、后的均值信号提取其中的本底信号,并采用模糊加权均值滤波器对本底信号进行去噪,去除与裂纹信号混淆的噪声,最后根据信号矩阵、稀疏的裂纹信号和去噪后的本底信号得到低信噪比的AE信号.试验结果表明,所提算法有效去除了AE信号中易与裂纹信号混淆的噪声,使故障特征更加明显,成功刻画了滑动轴承不同摩擦状态间的变化.%For extracting the relatively weak crack information contained in plain bearing Acoustic Emission (AE) signals,in cosideration of the signal sensibility of K-SVD algorithms,an improved average signal method based on the K-SVD dictionary was proposed.The sparse and pulse signal extraction characteristics of the AE signal matrix were obtained by using the signal reorganization and expansion strategy,which avoids the mixed noise pollution on the AE signal.Then,a fuzzy weighted average filter was applied to process the remained signal,which eliminates the mixed noise pollution on the low amplitude signals.The superimposition of the average signal in K-SVD was achieved by the above two steps.Compared with the traditional K-SVD algorithm,the improved algorithm can achieve better denoising performance and more obvious fault features.The experimental results show the change of the bearings friction state,which validates the effectiveness of the algorithms at the same time.
展开▼