【摘要】 在信号的记录和处理中,缺失数据导致信号不完整的情况常常存在。在记录心电信号时,由于种种原因,RR间期信号偶尔会出现连续的一段空白,即存在缺失数据。为了探究RR间期缺失数据的几种插值方法的优劣,本文提出并应用了一种用logistic混沌序列插值的新方法。采用年老组和年轻组各13名健康受试者的RR间期序列为实验数据,随机除去部分连续数据段,构造数据缺失序列。根据除去数据段的长短不同,构造不同缺失率的模拟数据。用线性插值、样条插值和Logistic混沌序列插值三种方法对缺失段进行模拟,分别计算原始数据和模拟数据的近似熵和样本熵两个指标,比较同一缺失率下三种插值方法的模拟效果。通过计算,得到结果:在相同的缺失率下,混沌序列可以较好的模拟缺失数据,插值200个点后的样本熵与原始值无明显变化(年老组p=0.3;年轻组p>0.2),而线性、样条模拟的样本熵与原始值相比有明显改变(p<<0.01)。最终我们有初步结论:在缺失率较小时,三种方法的模拟效果差异不大。随着缺失数据的增多,三种方法的模拟效果都有逐渐降低的趋势。三种方法的模拟效果与缺失率有关。在一定缺失率下,混沌序列插值方法优于线性插值和样条插值方法。更多还原

【Abstract】 Incomplete signals with missing data always exist in the records and processing of actual signals.When RR interval data is recorded,a small part of blank,i.e.a missing duration appears.To explore a better interpolation method of missing RR-interval data,logistic chaotic series is put forward and used in the paper.The incomplete data with missing data were obtained by removing some randomly selected data from the original data,which obtain RR interval series of the old and the young groups with 13 healthy individuals.The missing data duration with different rates was made because of the different number of randomly selected data.And the randomly removed data were considered as a section that need to be interpolated,using several interpolation methods(linear,spline and logistic chaotic series).We worked out the approximate and sample entropy and compare these methods.Result:The logistic chaotic series can simulate original data better after the Interpolation of 200 data,the Sample Entropy being almost the same with the original(the old group p=0.3;the young group p>0.2),while others having obvious difference(p<<0.01).We have the conclusion that the simulation effect of three methods are similar to each other,but with the increasing missing rate,and the simulation effect has presented a decreasing trend.It’s related to the number of missing data.The interpolation of the logistic chaotic series is better than another two methods,to some extent.更多还原