冗余提升多小波包的构造及其应用

针对提取隐藏在原始振动信号中的弱周期性冲击信号,提出结合冗余提升多小波包(RLSMWP)及滑动窗奇异值分解(SWSVD)降噪的方法。利用提升方案实现具有5阶逼近阶性质的冗余Haar预处理,对信号进行预滤波,获得2重矢量信号。对多小波分解得到的矢量细节信号进行进一步分解,实现冗余提升多小波包变换。对最后一层各输出通道信号进行SWSVD降噪,重构后获得降噪信号。结果表明,RLSMWP与SWSVD相结合具有很好的降噪效果,提取出了隐藏在气阀振动信号中的弱周期性冲击成分;与传统多小波构造方法相比,新方法在时域实现了预滤波、多小波分解、多小波重构及后处理,具有计算简单、节省内存、运算速度快、可完全重构等优点。

Construction of redundant lifting scheme multi-wavelet packets and its application

Aiming at the extraction of weak period impact components buried in original signal, a novel method to combine redundant lifting scheme multi-wavelet packets(RLSMWP) and sliding window singular value decomposition(SWSVD) was developed. Original signal was pre-filtered by redundant Haar preprocessing with five approximation order, and vector signal was calculated. The vector approximation signal was calculated by redundant updater and rescaled and decomposed furthermore.The redundant lifting scheme multi-wavelet packet transform was realized. Every output channel signal on the last level was denoised using SWSVD, then signal was reconstructed using reconstruction algorithm. The results show that the noise reduction effect is good using RLSMWP-SWSVD method, and the weak fault signal of a valve was extracted from the strong vibration background. Compared with the traditional multi-wavelet construction algorithm, the new method achieves prefiltering, multi-wavelet decomposition, multi-wavelet reconstruction and post-processing in the time domain. The new method is characterized by simple calculation and high computing speed. It can save memory and can be completely reconstructed.