中文摘要：

时间序列的符号动力学信息熵Hk因其计算简单快速,对数据量要求小,而被应用于心率变异性（heart rate variability,HRV）分析,然而符号化的参数选择至今却并未形成统一标准.HRV作为典型的生理信号,存在着极大的个体间差异和非平稳性,要获得稳健的一致性分析,在符号化过程中必须考虑符号化参数α与序列本身均值、标准差的综合影响.文中,首先以仿真噪声序列为对象,考察了3个参数对于Hk的影响及三者相互之间的关联性,研究表明当满足特定关系时,Hk的曲线簇收敛于反映序列动力特性的Hk-up;随后在对15例心跳间隔序列的分析中,验证了Hk-up在消除个体间差异及减弱非平稳干扰影响两方面都优于α取固定值时的研究结果.

英文摘要：

The Shannon entropy Hk in symbolic dynamics analysis of time series has less data demand and can be complemented easily,and therefore is applied to heart rate variability （HRV） analysis. However,the criterion has not been established as to the parameter α,which is used during the transformation into symbols. Like all other physiological signals,great differences between individuals as well as nonstationarity are usual in HRV. Therefore,to achieve a robust and consistent analysis,the parameter α should be considered together with the mean and standard error of the series. In this paper,the integrative effects on Hk of the three parameters were studied firstly in simulation time series,and it was suggested that under certain conditions,the clusters of Hk approach a convergent upper boundary named Hk-up,which reflects the intrinsic dynamics of the time series. Then,in the heart beat interval series analysis of 15 subjects,the abilities of Hk-up to alleviate the impacts brought by both difference between individuals and nonstationarity were testified.