论文部分内容阅读
针对旋转机械常处在强背景噪声中的特点,应用SVD 降噪不能准确选取重构阶次的问题,本文提出基于相关系数的时频矩阵SVD 降噪方法.首先对振动信号进行EEMD分解获得IMF 分量,通过对IMF 分量构建的时频矩阵进行奇异值分解,将振动信号的频率信息分解到不同的时频子空间中,计算各时频子空间自相关函数与原信号自相关函数的相关系数;然后,根据相关系数选取合适的时频子空间进行信号重构,从而实现振动信号降噪.对仿真信号和实测健康振动信号进行降噪试验,并与传统奇异值分解降噪进行对比,表明了该方法的可行性和有效性.“,”According to the problem of the reconstruction order cannot be selected accurately by singular value decomposition(SVD) and noise reduction, and rotating machinery is often in the strong background noise, a time frequency matrix SVD de-noising method based on correlation coefficients is proposed. Firstly, the vibration signal is decomposed by ensemble empirical mode decomposition (EEMD) and intrinsic mode function (IMF) component is obtained, time-frequency matrix that is constructed by IMF component could be decomposed by SVD, so that the frequency information of vibration signal is decomposed into different time-frequency subspaces. The correlation coefficients between time-frequency subspaces component autocorrelation function and vibration signal autocorrelation function are calculated. Second, appropriate time-frequency subspaces components are chosen to reconstruct signal based on correlation coefficients so as to realize the signal noise reduction. Through performing noise reduction test for simulated signals and measured healthy vibration signals, and comparing with the traditional singular value decomposition and noise reduction, it is proved that this method is feasible and effective.