摘要
将产生非高斯信号的非线性系统信息用高阶统计量中的双谱来表达,以检测非线性系统的二阶相位耦合信息,并用双相干系数定量估计二次相位耦合的耦合程度。通过对比分析典型的非高斯信号(EEG信号)在不同脑功能状态下的双谱结构,选取双相干系数的对角切片及与对角切片平行的其他切片作为特征提取对象,最终对每一切片元素采取求和处理形成有效双谱特征来构造特征向量,并用支持向量机分类器进行分类,取得了良好的效果。这表明用有效的双谱特征分析非高斯信号能更有效地提取有用信息。
The information of nonlinear system which has generated non-Gaussian signal was expressed by bispectrum,which can detect quadratic phase coupling information of a nonlinear system,and the quadratic phase coupling degree was quantitative estimated by bicoherence coefficient.The bispectrum structures of the typical non-Gaussian signals(EEG signals) in different brain function status were analyzed,and for the bicoherence coefficient,let its diagonal slice and other slices parallelling to the diagonal slice as the feature extraction objects.And then the sum processing was carried on each slice,so the feature vector was formed.Finally the classification accuracy achieved the significant improvement via adopting the support vector machine classifier.It shows that non-Gaussian signal is analyzed by the effective spectral characteristics can more efficiently extract useful information.
出处
《中南大学学报(自然科学版)》
EI
CAS
CSCD
北大核心
2013年第S1期306-309,共4页
Journal of Central South University:Science and Technology
基金
国家自然科学基金资助项目(61064003)
国家自然科学基金资助项目(51165024)
关键词
非线性系统
双谱
双相干系数
脑电
特征提取
nonlinearsystem
bispectrum
bicoherence coefficient
EEG
feature extraction