脉搏信号的生物识别

Biological Identification of Pulse Signal

以中医脉象人工智能辨识系统的研究与开发为背景,对脉搏信号的分析与识别进行了研究与探讨。首先对平静状态和兴奋状态下的脉搏波进行聚类,聚类后再进行脉搏波的生物识别。将脉搏波幅度序列看做非平稳时间序列,并对其进行解析,将其中的趋势分量和波动分量进行了分离,运用时变参数自回归(TVPAR)模型得到残差,最后采用马氏距离判别法进行判别分析,得到了比较理想的识别结果。

The pulse analysis is researched on the base of artificial intelligence system of Chinese traditional pulse diagnosis. Firstly, the pulse of calmness and excitement conditions are clustered, then the pulse is identified. The amplitude series of the pulse signal are analyzed as non-stationary time series, of which the trend component and fluctuant one are separated. The error item is obtained by the time varying parameter autoregressive （TVPAR） model. Finally, according to Mahalanobis distance, pulse identification experiment is carried out.