DTS-AMS型神经网络用于自发脑电信号残部的外推估计──提取视觉诱发脑电信号的新方法之二

Extrapolation of the Electroencephalogram Residual Via DTS-AMS Type Neural Network

摘要: 利用可实现对任意阶多元多项式超曲面的无差逼近的、基于离散泰勒级数的高阶联想记忆系统(DTS-AMS),对(本系列文章之一中所得)自发脑电(EEG)信号错相叠加平均后的残余信号进行建模和外推估计.数字仿真结果表明,此法通过十次左右的视觉刺激,即可将视觉诱发脑电(VEP)信号从EEG信号中分离出来,所得VEP信号潜伏期P₁₀₀精度可满足一般临床诊断要求.

Abstract: Utilizings the higher-oder associative memory system via discrete Talor series (DTS-AMS) which can make an error-free approximation to the multi-variable polynomial function with arbitrary order, this paper realizes the modeling and extrapolation of the EIectroencephalogram(EEG) residual signal obtained from the phase-shifting and superposition. Numerical simulation results have shown that, through about only 10 times of visual stimulation, the Visual Evoked Potential(VEP) can be extracted from the EEG background with a satisfactory identification accuracy of the latency parameter P100 for the clinical use.