摘要
近年来的研究表明,采用α稳定分布模型来描述带噪EP信号比用常规的高斯模型具有更好的适应性。在α稳定分布噪声下,特别是当噪声的非高斯特性比较明显时,常规的基于二阶或高阶统计量的算法其性能将有所退化,甚至不能工作。本研究提出基于线性观测模型的数据复用最小平均p范数(DR-LMP)方法,并将之用于α稳定分布噪声下单路EP信号的动态提取。仿真结果表明,此方法能够在一定程度上减少提取EP信号时所需的刺激实验次数,实现单导EP信号的少次提取。当观测EP信号的信噪比大于-10dB时,此方法能够有效跟踪EP信号的变化,实现EP信号的动态提取。此外,本方法改变了二阶算法在分数低阶噪声下不收敛的缺点,在高斯噪声和分数低阶噪声下均具有良好韧性。
Recently it is accepted that Alpha stable distribution is better for modeling highly noise contaminated EPs than Gaussian distribution.When algorithms based on the second or higher order statistics are used for estimating EPs with Alpha stable distribution noise,the performance of them degenerates.This paper presented a data reuse least mean p-norm algorithm for the dynamic estimation of single channel EPs.The simulation results showed that the proposed algorithm could extract EP signals with less stimulation trails,and dynamically estimate single channel EP signals（SNR〉-10dB）.In addition,the algorithm converged well in Alpha stable noise conditions and was more robust than conventional algorithms.
出处
《中国生物医学工程学报》
CAS
CSCD
北大核心
2008年第5期684-688,共5页
Chinese Journal of Biomedical Engineering
基金
国家自然科学基金资助项目(30570475,60772037)
教育部博士点基金(20050141025)
关键词
线性观测模型
数据复用最小平均p范数算法
诱发电位
linear observation model
data reuse least mean p-norm algorithm（DRLMP）
evoked potential（EP）
