摘要
目前,大多数的人体睡眠状态辨识方法都是采用幅值域、频域或时-频域的线性分析方法,来处理脑电信号这一含有大量非线性成分的非平稳随机信号,因而其辨识效果的鲁棒性和稳定性无法保证.为此,本文提出了一种非线性的时-频辨识方法.该方法对脑电信号进行重排伪Wigner-Ville分布的时-频分析,并利用Wigner-Ville分布边缘聚集特性在不同频段内提取特征量,最后通过支持向量机实现睡眠状态的精确辨识.实验表明,该方法辨识睡眠状态的准确率达到91.6%,鲁棒性得到显著改善,为后续进行睡眠控制研究奠定了基础.
At present,most human sleep state recognition method based on time-frequency domain amplitude domain,frequency domain and the linear analysis method,processing EEG signal that contains a lot of nonlinear components of non-stationary random signal,thus recognition effect of robustness and stability cannot be guaranteed.To this end,this paper proposes a nonlinear identification of time-frequency method,the method of brain electric signal rearrangement pseudo Wigner Ville distribution-time-frequency analysis,and use the Wigner Ville distribution-gathered edge features in different frequency bands to extract the characteristic,finally,support vector machine(SVM)to realize precise identification of sleep.The results showed that the method recognized the accuracy of 91.6% and improved robustness,and laid the foundation for follow-up sleep control study.
引文
[1]金永寿.健脑,更要合理用脑[J].家庭医药,2014(4):24.
[2]周宏,李玉斌.睡眠对人生的重要作用[J].中国社区医师,2002,18(14):11.
[3]李谷,范影乐,李轶,等.基于脑电信号Hilbert-Huang变换的睡眠分期研究[J].航天医学与医学工程,2007,20(6):458-463.
[4]刘慧,谢洪波,和卫星,等.基于模糊熵的脑电睡眠分期特征提取与分类[J].数据采集与处理,2010,25(4):484-489.
[5]Diykh M,Li Y,Wen P.EEG sleep stages classification based on time domain features and structural graph similarity[J].IEEE Transactions on Neural Systems&Rehabilitation Engineering,2016,24(11):1 159-1 168.
[6]周强,陈颖,李俊雨,等.基于EEG信号的STPS睡眠状态在线辨识[J].陕西科技大学学报(自然科学版),2016,34(6):164-170.
[7]张贤达,保铮.非平稳信号分析与处理[M].北京:国防工业出版社,1998.
[8]杨贵琳.基于生物互感原理的粉红噪声睡眠仪的研究[D].西安:陕西科技大学,2014.
[9]Kawada T,Suzuki S,Aoki S,et al.Effects of noise on sleep.Part 2.A case report of the effects of three levels of stationary sound on sleep parameters[J].Nihon Eiseigaku Zasshi.Japanese Journal of Hygiene,1989,43(6):1 102-1 108.
[10]周强.基于噪声分析的造纸软测量理论方法研究和应用[D].西安:西安交通大学,2009.
[11]Wu X,Liu T.Spectral decomposition of seismic data with reassigned smoothed pseudo Wigner-Ville distribution[J].Journal of Applied Geophysics,2009,68(3):386-393.
[12]Liu Dehua,Qian Hui,Dai Guang,et al.An iterative SVM approach to feature selection and classification in high-dimensional datasets[J].Pattern Recognition,2013,46(9):2 531-2 537.
[13]李钢,王蔚,张胜.支持向量机在脑电信号分类中的应用[J].计算机应用,2006,26(6):1 431-1 433.