A precise feature extraction method for shock wave signal with improved CEEMD-HHT
A precise feature extraction method for shock wave signal with improved CEEMD-HHT
复制标题
DOI：
10.1007/s12652-020-02204-7
复制
发表时间：
2020-06
影响因子：
--
通讯作者：
Zonglei Mou;Xueben Niu;Chen Wang
中科院分区：
文献类型：
--
作者：
Zonglei Mou;Xueben Niu;Chen Wang
有效提取特征参数是评估武器毁伤性能的关键。目前，许多经典的特征提取算法都存在提取结果不能满足实际需要的问题。针对复杂测试环境下冲击波超压信号噪声大、特征提取困难等问题，提出了一种基于改进互补系综经验模态分解（CEEMD）和Hilbert-Huang变换（HHT）的精确特征提取方法。引入CEEMD对原始爆炸冲击波信号进行分解，并采用小波包阈值降噪方法从噪声高的高频内禀模态函数中提取有用信息。引入相关系数算法去除不相关的imf。此外，我们对原始信号进行重构，提取真实的时间过程特征，并利用Hilbert-Huang变换（HHT）精确提取各种imf的瞬时特征和能谱。改进的CEEMD-HHT是一种精确的激波信号分析方法。它既能有效地去除噪声，又能保留有效的高频信息而不丢失有用信息。此外，该方法克服了经验模态分解（EMD）中模态混合的问题，具有特征提取精度高、自适应等优点。通过两组实验数据验证了该方法的有效性，并能准确提取冲击波超压信号的瞬时特征和能量谱，为武器毁伤评估提供了新的理论依据。
Efficient extraction of feature parameters is the key to evaluating weapon damage performance. At present, many classical feature extraction algorithms have the problem that the extraction cannot meet the actual needs. A precise feature extraction method based on improved complementary ensemble empirical mode decomposition (CEEMD) with Hilbert-Huang Transform (HHT) was proposed in this paper to solve problems such as large noise and difficulties in extracting features of shockwave overpressure signals in complex test environment. We introduced CEEMD to decompose original explosion shockwave signals and adopted wavelet packet threshold de-noising to extract useful information from noisy high-frequency intrinsic mode functions (IMFs). The correlation coefficient algorithm is introduced to remove unrelated IMFs. In addition, we performed reconstruction of original signals to extract true time-course feature and utilized Hilbert-Huang Transform (HHT) to achieve precise extraction of instantaneous feature and energy spectrum of the various IMFs. The improved CEEMD-HHT is a precise method for shock wave signal analysis. It not only effectively removes noise, but also retains effective high-frequency information without losing useful information. Additionally, it overcomes the problems of mode mixing in empirical mode decomposition (EMD), and has the advantages of feature extraction with high accuracy and self-adaptation. The effectiveness of the proposed method is demonstrated by 2 groups of experimental data, and it precisely extracts instantaneous feature and energy spectrum of shockwave overpressure signal, which provide new theoretical basis for the evaluation of weapon damage.