电子测量技术ELECTRONICMEASUREMENTTECHNOLOGY第46卷第1期2023年1月DOI:10.19651/j.cnki.emt.2210358基于改进扩展典型相关分析的SSVEP信号识别方法*芦鹏戴凤智尹迪温浩康高一婷(天津科技大学电子信息与自动化学院天津300222)摘要:现有的稳态视觉诱发电位(SSVEP)的信号识别方法没有充分关注信号的相位特征在识别过程中的重要作用,为此提出一种扩展典型相关分析(eCCA)的改进方法。将联合频率-相位调制编码的刺激范式中的相位参数添加到由受试者训练数据所构造的参考信号,以此来实现对eCCA的相位约束,从而提升eCCA方法对SSVEP信号的识别性能。通过在公开数据集上与现有的SSVEP信号识别方法进行对比实验,表明所提方法对SSVEP信号的平均识别率提高到82.76%,信息传输速率提高至116.18bits/min,且具有更好的稳定性。关键词:稳态视觉诱发电位;脑机接口;脑电信号;扩展典型相关分析中图分类号:TN911文献标识码:A国家标准学科分类代码:510.40SSVEPsignalidentificationmethodbasedonimprovedextendedcanonicalcorrelationanalysisLuPengDaiFengzhiYinDiWenHaokangGaoYiting(CollegeofElectronicInformationandAutomation,TianjinUniversityofScienceandTechnology,Tianjin300222,China)Abstract:Manyexistingsignalrecognitionmethodsforsteady-statevisualevokedpotential(SSVEP)donotpaysufficientattentiontotheimportanceofthephasefeatures.Inthispaper,animprovedextendedcanonicalcorrelationanalysis(eCCA)methodisproposedforSSVEPsignalidentification.Thephaseparameterinthestimulusparadigmofjointfrequency-phasemodulationcodingisaddedtothereferencesignalconstructedfromsubjects'trainingdataasawaytoachievephaseconstraintsoneCCA,thusimprovingtherecognitionperformanceoftheeCCAmethodforSSVEPsignals.ThustheeCCA-basedSSVEPsignalrecognitionperformanceisimproved.Toverifytheeffectivenessoftheproposedmethod,SSVEPsignalrecognitionexperimentsareconductedonapubliclyavailabledatasetandcomparedwiththeexistingsignalrecognitionmethods.Theexperimentalresultsshowthattheaveragerecognitionrateoftheproposedmethodisimprovedto82.76%,andtheinformationtransmissionrateisreachedto116.18bits/minwithbetterstability.Keywords:steady-statevisualevokedpotential(SSVEP);braincomputerinterface(BCI);electroencephalogr...
