电子设计工程
電子設計工程
전자설계공정
ELECTRONIC DESIGN ENGINEERING
2015年
8期
42-45,48
,共5页
脑电%双谱%分类%支持向量机
腦電%雙譜%分類%支持嚮量機
뇌전%쌍보%분류%지지향량궤
electroencephalogram%bispectrum%classification%support vector machine
脑电是一种典型的非高斯、非线性随机信号，传统时域或频域分析已不能准确表征信号特征，而高阶谱方法对脑电信号的处理却有较好效果。本文通过几组不同意识状态下的脑电测试实验，提取脑电信号，并通过双谱分析提取脑电双谱切片的特征值，借助支持向量机、概率神经网络、最近邻分类算法等3种方法对双谱切片的特征值进行处理，比较其分类效果。研究结果表明，电极C3处脑电与其他电极处的脑电具有不同的双谱特征，不同脑电极信号双谱切片具有明显差异。
腦電是一種典型的非高斯、非線性隨機信號，傳統時域或頻域分析已不能準確錶徵信號特徵，而高階譜方法對腦電信號的處理卻有較好效果。本文通過幾組不同意識狀態下的腦電測試實驗，提取腦電信號，併通過雙譜分析提取腦電雙譜切片的特徵值，藉助支持嚮量機、概率神經網絡、最近鄰分類算法等3種方法對雙譜切片的特徵值進行處理，比較其分類效果。研究結果錶明，電極C3處腦電與其他電極處的腦電具有不同的雙譜特徵，不同腦電極信號雙譜切片具有明顯差異。
뇌전시일충전형적비고사、비선성수궤신호，전통시역혹빈역분석이불능준학표정신호특정，이고계보방법대뇌전신호적처리각유교호효과。본문통과궤조불동의식상태하적뇌전측시실험，제취뇌전신호，병통과쌍보분석제취뇌전쌍보절편적특정치，차조지지향량궤、개솔신경망락、최근린분류산법등3충방법대쌍보절편적특정치진행처리，비교기분류효과。연구결과표명，전겁C3처뇌전여기타전겁처적뇌전구유불동적쌍보특정，불동뇌전겁신호쌍보절편구유명현차이。
Electroencephalogram (EEG) is a very complex non-Gaussian, nonlinear stochastic signal, using the time domain or frequency domain analysis can not accurately characterize the characteristics, but the high-order spectrum analysis methods for the use of EEG signal processing research does have a good effect. In this paper, by doing several different groups of mental EEG experiment, we obtained the EEG bispectrum slice of every experimental data, extracted the characteristic values of the bispectrum slice, and used SVM, KNN, PNN classifications methods to process the spectral slice. Then the classification accuracy was obtained in different experimental conditions. The result shows that SVM classifier has the highest classification accuracy; compared C3 electrode with other 11 electrodes, the signals of different slice of brain electrode have different bispectral characteristics.
