摘要
希尔伯特-黄变换(Hilbert-Huang Transform,简称HHT)方法是一种自适应性信号处理方法,在处理非线性、非稳态信号方面有很大优势。但HHT分解复杂信号时存在求解结果精确不高、计算时间长等不足。针对HHT的边端效应、越界问题、停止准则和虚假低频成分过滤等问题,文章提出了相应的改进方法。为有效抑制边端效应,人为定义两个极值点,然后连接相邻极值点形成直线后平行延拓。利用信号与包络线的极限差值多次拟合包络线,初步解决了越界问题。根据虚假成分与原始信号的相关系数远小于真实信号与原始信号的相关系数,成功过滤掉虚假成分。数值算例的结果表明了所提方法的有效性。
Aim.The HHT method has great advantages in processing non-linear and non-steady-state signals.But,when the signals are complex,there are,in our opinion,some disadvantages pointed out in the introduction of the full paper.Section 1,2,3 explain our improvements in the following three aspects： end effect problem,overshoot problem,sifting stop criteria and filtering of false low-frequency components.Section 1 improves the HHT method by defining two extreme value points,connecting adjacent maximum extreme value point and minimum extreme value point,and by fitting the envelope line of the complex signal to be decomposed.Section 2 first finds the overshoot points,which have the maximum difference between the original signal and the envelope signal,and then takes these overshoot points as maximum values to carry out the cubic spline refitting of the envelope line thrice.Section 3 gives eqs.（2） and（3） to obtain the time for stopping the signal decomposition when the correlation coefficient of a certain order IMF signal becomes obviously small,as shown in Table 1.To verify the effectiveness of our improvements,section 4 gives a numerical example for analyzing the response of a composite shell and detecting its small damage.The numerical results,gives in Table 2,3 and 4,and their analysis show preliminarily that our improved HHT method can effectively decompose IMF signals and detect small structural damage.
出处
《西北工业大学学报》
EI
CAS
CSCD
北大核心
2011年第2期268-272,共5页
Journal of Northwestern Polytechnical University
基金
国家自然科学基金(11072197)
西北工业大学基础研究基金(JC201033)资助
关键词
HHT
边端效应
越界问题
停止准则
虚假低频成分
signal processing
monitoring
structural damage monitoring
Hilbert-Huang transform（HHT）
end effect problem
overshoot problem
sifting stop criterion
false low-frequency component
