基于经验模态分解的脉搏信号特征研究

Pulse signal feature research based on empirical mode decomposition

通过研究了脉搏信号的时域特征,利用阈值法提取了脉搏信号的时域特征点,得到了潮波与主波峰值的比值以及脉图K值。在分析经验模态分解算法的基础上,利用其对非稳态信号分解的自适应性,对典型高血压病人和正常人脉搏信号进行了EMD分解,研究了脉搏信号各模态的能量特征,提出了模态能量商的概念。对临床采集的130例高血压病人、高校50例健康中年人和50例健康大学生脉搏信号分别进行潮波与主波比值、K值和模态能量商特征计算,结果表明模态能量商能够区分正常人和高血压病人,并且与高血压弦脉程度有较好的相关性。实验证明所提出的模态能量商特征具有较好的重复性与稳定性,可以作为脉搏信号识别的特征向量。

The time domain features of pulse signal are studied and are extracted using threshold method in this paper. The predicrotic to main wave ratio and the K value of the pulse wave are obtained. Based on the algorithm of Empirical Mode Decomposition （EMD）, which can adaptively decompose the non-stationary signal, the pulse signals of typical patients with hypertension and normal persons are decomposed. The energy feature of each mode is studied and the concept of mode energy ratio is proposed. The predicrotic to main wave ratio, the K value and the mode energy ratio of the pulse signals , which were collected from 130 cases of clinic hypertension patients, 50 cases of middle-aged healthy persons and 50 cases of healthy students in university, are respectively calculated. Results show that the mode energy ratio is able to distinguish between normal persons and patients with hypertension, and the mode energy ratio has good correlation with the degree of wiry pulse in hypertension patients. Experiments testify that the energy ratio has better repeatability and stability, and can be regarded as a feature vector to identify the pulse signals.