单通道脑电信号疲劳分类检测  

Fatigue Classification Detection of Single-Channel EEG Signal

作　　者：张怀勇 张振昌[1] ZHANG Huaiyong;ZHANG Zhenchang(College of Computer and Information,Fujian Agriculture and Forestry University,Fuzhou,Fujian 350002,China)

机构地区：[1]福建农林大学计算机与信息学院,福建福州350002

出　　处：《闽江学院学报》2024年第5期31-41,共11页Journal of Minjiang University

基　　金：福建省自然科学基金项目(2021J01124)。

摘　　要：为解决头皮脑电信号电位信号微弱、易受干扰、具有非平稳性和随机性、手动提取特难度大的问题,提出了CNN+BiGRU的网络模型,充分提取了EEG信号前后序列之间的关联信息。实验结果表明,提出双流网络模型准确率达到95.70%。与现有的研究方法相比,显著提高了单通道脑电信号进行疲劳检测的准确性与可行性,为疲劳检测研究提供了新思路。To solve the problem of weak scalp EEG potential signals,susceptibility to interference,non stationarity and randomness,and difficulty in manually extracting features,we propose a CNN+BiGRU network model that fully extracts the correlation information between EEG signal sequences before and after.The experimental results show that our proposed dual stream network model achieves an accuracy of 95.70%.Compared with existing research methods,the accuracy and feasibility of fatigue detection using single channel EEG signals have been significantly improved,providing new ideas for fatigue detection research.

关 键 词：脑电信号 深度学习 BiGRU 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]