基于残差卷积脉冲神经网络的稳态视觉诱发电位脑电信号研究  

Research on Steady State Visual Evoked Potential EEG Signals Based on Residual Convolutional Spiking Neural Network

作　　者：王刚 高福临 孙凯明 郝明 周丽丽 WANG Gang;GAO Fu-lin;SUN Kai-ming;HAO Ming;ZHOU Li-li(Institute of Intelligent Manufacture Heilongjiang Academy of Science,Harbin 150090 China)

机构地区：[1]黑龙江省科学院智能制造研究所,黑龙江哈尔滨150090

出　　处：《自动化技术与应用》2024年第11期182-186,共5页Techniques of Automation and Applications

基　　金：黑龙江省省属科研院所科研业务费项目(CZKYF2023-1-C032)。

摘　　要：随着脑机接口技术的飞速发展以及类脑人工智能在脑信号识别领域的深入探索,本文设计一种基于残差卷积脉冲神经网络(ReC-SNN)的稳态视觉诱发电位(Steady State Visual Evoked Potentials,SSVEP)脑电信号(Electroencephalogram,EEG)识别方法,解决稳态视觉诱发电位脑电信号识别率低问题。首先通过对SSVEP脑电信号获得、预处理构建数据集,其次设计ReC-SNN结构,并优化网络参数,进而实现对字母、数字、功能键图片在对应频率刺激产生的脑电信号识别,其识别精度达到95%以上。实验结果表明,本文设计的ReC-SNN网络模型能够有效地识别SSVEP脑电信号,为类脑智能识别脑电信号技术的发展提供了支撑。With the rapid development of brain-computer interface technology and the in-depth exploration of brain-inspired intelligence in the field of brain signal recognition,the paper designs a SSVEP(Steady State Visual Evoked Potentials)EEG(Electroencephalogram)signal recognition method based on Residual Convolutional Spiking Neural Network(ReC-SNN).The network solves the low recognition rate of Steady State Visual Evoked Potential EEG signal.Firstly,the dataset is constructed by obtaining and pre-processing the SSVEP EEG signals.Secondly,the dataset is applied to train and optimise the network parameters by designing ReC-SNN structure.In turn,the precise recognition of EEG signals generated by stimulation of letters,numbers,and pictures of function keys at corresponding frequencies is achieved.Its recognition accuracy reaches more than 95 percent.The experimental results show that the ReC-SNN network model designed in this paper can effectively recognise SSVEP EEG signals,which provides support for the development of Brain-inspired Intelligence recognition EEG signal technology.

关 键 词：脑电信号 SSVEP ReC-SNN 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]