联合经验模式分解和混沌理论的稳态视觉诱发电位脑电识别  

Combined Empirical Pattern Decomposition and Chaos Theory for Steady-State Visual Evoked Potential Electroencephalogram Recognition

作　　者：郭晓冰 徐光华[1,2,3] 李辉 谢杰仁[1] 江翰立 张四聪 GUO Xiaobing;XU Guanghua;LI Hui;XIE Jieren;JIANG Hanli;ZHANG Sicong(School of Mechanical Engineering,Xi’an Jiaotong University,Xi’an 710049,China;State Key Laboratory for Manufacturing Systems Engineering,Xi’an Jiaotong University,Xi’an 710049,China;The First Affiliated Hospital of Xi’an Jiaotong University,Xi’an 710061,China)

机构地区：[1]西安交通大学机械工程学院,西安710049 [2]西安交通大学精密微纳制造技术全国重点实验室,西安710049 [3]西安交通大学第一附属医院,西安710061

出　　处：《西安交通大学学报》2024年第6期34-42,共9页Journal of Xi'an Jiaotong University

基　　金：科技创新2030基金资助项目(2021ZD0204300,2022ZD0209800);广州市科技计划资助项目(202206060003)

摘　　要：针对多目标稳态视觉诱发电位(SSVEP)信号识别准确率低、线性识别方法抑制噪声的同时也会抑制信号本身特征等问题,基于经验模式分解对信号的降噪特性、达芬混沌系统对微弱周期信号的敏感性及噪声的免疫特性,提出了一种非线性多目标SSVEP信号识别算法。首先,采用共平均参考算法将多通道SSVEP信号融合成单通道信号,通过傅里叶变换求得SSVEP信号的相位谱,为达芬混沌系统周期策动力添加相位;接着,采用经验模式分解降噪,将获得的第一个本征模函数输入到达芬混沌系统中,利用基于频谱差异的混沌系统状态判别方法,求解各目标的刺激频率幅值;最后,根据最大刺激频率幅值确定刺激目标,实现了对多目标SSVEP信号的识别。研究结果表明:相较于典型相关分析法,所提非线性信号处理方法的平均识别准确率提高了7.3%,平均信息传输速率提高了3.84bit/min。该研究为探究非线性SSVEP信号解码算法提供了新方向。Aiming at the problems of low recognition accuracy of multi-target SSVEP signals and suppression of the characteristics of signal while the linear recognition method suppresses noise,this paper proposes a nonlinear multi-objective SSVEP signal recognition algorithm based on the noise reduction characteristics of empirical mode decomposition,Duffing chaotic system’s sensitivity to weak periodic signals and the immune characteristics of noise.First,multi-channel SSVEP signals are integrated into single-channel signals by using the co-average reference algorithm.Then,the phase spectrum of SSVEP signal is obtained by Fourier transform,and the phase is added to the periodic dynamics of Duffing chaotic system.After empirical mode decomposition noise reduction,the first eigenmode function obtained is input into Duffing chaotic system,and the amplitude of each target stimulus frequency is solved by using the chaotic system state discrimination method based on spectral difference.Finally,the stimulus target was determined according to the amplitude of the maximum stimulus frequency and multi-target SSVEP signals are recognized.The results show that the average recognition accuracy of the proposed nonlinear signal processing method is 7.3%higher than that of typical correlation analysis.The information transmission rate increases by 3.84 bit/min.The method provides a new direction for exploring nonlinear SSVEP signal decoding algorithms.

关 键 词：稳态视觉诱发电位 达芬混沌系统 非线性信号处理 经验模式分解 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]