论文部分内容阅读
针对传统经验模态分解算法存在的端点效应问题,提出了一种适用于脉搏信号分析的基于模板匹配和镜像延拓的两阶段经验模态分解算法。依据脉搏信号的类周期特征,首先识别其特征信息,分离出单一心动周期内的脉搏信号;然后使用信号相干平均技术获取脉搏信号模板,依据模板将首尾端点处的脉搏信号扩展至整个心动周期;最后采用镜像延拓方法对扩展后的脉搏信号进行经验模态分解。实验结果表明,扩展后的脉搏信号能够较好地模拟原信号首尾端点处的变化趋势,因此新算法能够有效抑制传统经验模态方法存在的端点效应问题,适用于诸如脉搏信号等具有类周期特征的生理信号分析。
Aiming at the problem of the endpoint effect of traditional empirical mode decomposition algorithm, a two-stage empirical mode decomposition algorithm based on template matching and mirror extension is proposed for pulse signal analysis. According to the quasi-periodic characteristics of the pulse signal, the pulse signal is first identified by its characteristic information and the pulse signal is separated from the single cardiac cycle. Then the pulse signal template is obtained by using the signal coherence averaging technique and the pulse signal at the end of the pulse is extended to the whole cardiac cycle. Finally, the mirror extension method is used to decompose the expanded pulse signal under the empirical mode. The experimental results show that the extended pulse signal can simulate the trend of the end of the original signal, so the new algorithm can effectively restrain the end effect problem of the traditional empirical mode method. Physiological signal analysis.