基于噪声水平自适应估计的往复压缩机振动信号局部投影降噪方法

局部投影降噪算法在其应用过程中,邻域的选择对降噪效果有较大影响。提出了改进算法以解决传统算法中邻域难以选取的问题。该方法利用小波包分解技术,依据频带能量的差异将原始信号分解为噪声频带和系统信号频带,将噪声频带能量占原始信号能量的比值估计为噪声水平。在一定程度内逐步增加分解层数,直至该噪声水平收敛。根据收敛时的噪声水平估计相空间中相点的邻域半径,此外利用该噪声水平可实现对原始信号的盲信噪比估计。对含噪的Lorenz和Rossler序列进行数值仿真,结果表明该方法的降噪效果优于一些传统方法和基于定量递归分析的局部投影降噪算法。对实测往复压缩机振动信号的降噪研究,进一步表明了该方法的有效性。

Local projection noise reduction method for a reciprocating compressor vibration signal based on adaptive estimation of noise level

In practical application, selection of neighborhood radius has an important effect on performance of local projection noise reduction method.An improved method was proposed to solve the difficult problem of selection of neighborhood radius.Using this method, according to the difference of frequency band energy distribution, a vibration signal was decomposed into the noise frequency band and the system signal frequency band with wavelet packet method.The ratio of the energy of noise frequency band to the energy of vibration signal was estimated to be the noise level of vibration signal.In certain extent, the decomposition layers were increased gradually until the noise level converged.The neighborhood radius of the reconstructed phase space could be estimated using the convergent noise level.In addition, the blind SNR of the original vibration signal could be estimated using the convergent noise level.The numerical simulations of noisy Lorenz and Rossler series showed that the noise reduction effect of this method is better than those of some traditional methods and the local projection method based on recurrence quantification analysis.Noise reduction of a reciprocating compressor vibration signal further verified the effectiveness of this method.