基于小波包分解与主流形识别的非线性降噪

为解决工程实际中强噪声、非线性且频率成分复杂的振动信号降噪问题,提出了基于小波包分解和主流形识别的非线性降噪方法。采用小波包分解将原始振动信号正交无遗漏地分解到各频带范围内,根据各子频带中信噪空间分布,分别采用相应参数对小波包分解系数进行相空间重构;采用局部切空间排列(local tangent space alignment,LTSA)主流形识别方法在高维相空间中实现信号与噪音的分离,并重构出降噪后的一维小波包分解系数,最后进行小波包分解重构得到降噪后的振动信号。通过仿真实验和实例应用对本文所提方法的有效性进行了验证,试验结果表明本文方法具有良好的非线性降噪能力。

Nonlinear noise reduction method based on wavelet packet decomposition and principle manifold learning

Nonlinear noise reduction method based on wavelet packet decomposition and principle manifold learning is proposed, aiming to reduce the nonlinear noise of vibration signals with complex components in the practical engineering application. Firstly, the collected vibration signals are orthogonally decomposed into several sub frequency bands by wavelet packet decomposition, and the wavelet packet decomposition coefficients are reconstructed into a high dimensional phase space. Here, the parameters of phase reconstruction are selected according to the distribution of signal and noise in each sub frequency band and then principle manifold learning by local tangent space alignment (LTSA) is performed to separate the signal and noise in high dimensional phase space. Wavelet packet decomposition coefficients are reconstructed back into one dimensional series. At last, the vibration signal after noise reduction is obtained by the wavelet packet reconstruction. The effectiveness of the proposed method is simulated. The experiment and practical application, and the experimental results demonstrate the excellent nonlinear noise reduction capacity of the proposed method.