基于变模式分解的爆震特征识别方法

基于经验模态分解(empirical mode decomposition,简称EMD)算法因递归分解模式所造成的固有缺陷,将使用变分原理进行分解的变模式分解(variational mode decomposition,简称VMD)算法引入到爆震识别领域,发现VMD算法对比EMD算法有较高的计算效率与准确性,而且表现出了较好的鲁棒性,更加适合于在混有强烈背景噪声的缸盖振动信号中提取爆震特征。在此基础上,针对VMD算法分解层数需要手动选择的缺点,利用各阶分量的中心频率之差,提出了一种可以自适应选择VMD分解层数的方法。这种方法的思路为利用VMD算法对信号从一个较小的层数开始进行分解,逐个增加分解层数,直至各阶分量中心频率差值满足预先设定的阈值为止,即可得到最佳分解结果。经实验数据验证与对比,结果显示了这种方法的优越性。

Knock Detection Using Variational Mode Decomposition

The empirical mode decomposition (EMD) method has inherent defects because of a recursive decomposition. This paper introduces variational mode decomposition (VMD) into knock detection field, based on variational principle. Compared with the EMD, the VMD has better efficiency and accuracy, and more robust, which is better for knock detection in the vibration signal with strong background noise. In this case, this paper proposes an adaptive selection of VMD''s level number using the center frequency of different components, because the VMD method needs presetting the numbers of modal components. Decomposing a signal by the VMD in a low level, and increasing the decomposition level one by one are available till the center frequency of different components meet the predefined threshold, in whichthe best decomposition results can be obtained. The method is proved by the verification and comparison of experimental data.