基于AR模型的人体脉象信号模糊聚类研究

根据一种新的模糊聚类方法－F－PFSR（Fuzzy Pseudo F-Statistic Ratio)聚类法，对人体脉搏声信号（脉象信号）进行了聚类识别研究。首先对脉搏声信号作8阶AR模型拟合，模型系数构成信号的特征集，其次采用K－L变换对特征集进行了压缩，最后对临床实测脉象信号进行了聚类分析。实验结果表明，本文的聚类方法是可行和有效的。

Fuzzy clustering of human body pulse signals based on AR model

The human body pulse signals are clustered on the basis of a new kind of fuzzy clustering method - F-PFSR Clustering Method. Firstly the pulse signals are processed with AR model, the coefficients of the AR model constituting the characteristic sets of the signals. Secondly the characteristic sets are compressed by K-L transform. Lastly the pulse signals measured in clinic are clustered and analyzed. Experimental results show that our clustering method is effective.