基于CCA融合FFT的SSVEP脑机接口分类算法

A classification algorithm of a SSVEP brain-computer interface based on CCA fusion FFT

为解决多目标刺激范式的稳态视觉诱发电位脑电信号识别准确率低和信息传输率低的问题,提出了一种快速傅里叶变换同典型相关分析相结合的方法,通过快速傅里叶变换将信号训练成对应频率的训练模板,并作为参考信号与实时采集的信号进行典型相关分析来计算频率的识别准确率。6名受试者参与并完成了180组实验,在时间窗口长度为1.5 s的条件下,基于快速傅里叶变换-典型相关分析的稳态视觉诱发电位信号识别算法的平均识别准确率为93.98%,比典型相关分析算法提升了14.75%,信息传输率为62.30 bit·min^(-1),比典型相关分析算法提升了55.63%。实验结果表明,快速傅里叶变换-典型相关分析算法性能更优。

In order to solve the problems of low recognition accuracy and low information transmission rate of SSVEP EEG signals with multi-target stimulation paradigm,a method combining fast Fourier transform and typical correlation analysis was proposed using fast Fourier transform to train the signal into a training model corresponding to the frequency,and which was used as a reference signal to perform typical correlation analysis with the real-time acquired signal to calculate the recognition accuracy of the frequency.Six subjects participated and completed 180 sets of experiments.Under the condition of time window length of 1.5 s,the average recognition accuracy of the FFT-CCA-based SSVEP signal recognition algorithm was 93.98%,which was 14.75%better than the CCA algorithm,and the information transmission rate was 62.30 bit·min^(-1),which was 55.63%better than the CCA algorithm.The experimental results showed that the FFT-CCA algorithm has better performance and has good application prospects.