【摘要】 针对单电极可穿戴式脑电仪的脑电波信号(EEG)的疲劳状态智能识别,进行了基于广义回归神经网络(GRNN)的疲劳状态检测的研究。首先,通过调查问卷调查用户主观疲劳量,结合疲劳检测手环实现EEG数据的疲劳等级标记以建立数据集;其次,采用数据清洗等方式实现数据预处理并提取数据的时域特征、频域特征;运用主元分析进行特征降维;然后,建立GRNN疲劳识别模型并计算识别准确率;同时以支持向量机(SVM)方法作为对比实验检验模型效果;最后,以建立好的GRNN模型进行疲劳检测。研究发现,GRNN模型下EEG疲劳状态识别准确率最高为88.1%,相比SVM模型更高,对于EEG的疲劳状态的检测具有更好的稳定性和区分度。更多还原

【Abstract】 A study is carried out based on the fatigue state detection of general regression neural network(GRNN) for intelligent identification of the fatigue state of electroencephalogram(EEG) of a single electrode wearable electroencephalograph. First of all, the data sets are built by using the questionnaire to investigate the Karolinska sleepiness scale that the users feel about and the fatigue level marking EEG data based on fatigue detection bracelet. Then, the data preprocessing is realized by means of datacleaning in order to extract features from both time domain and frequency domain; principal component analysis is used to reduce the dimension of the data; the fatigue recognition model is established by GRNN and the recognition accuracy is calculated; and the support vector machine(SVM) method is used as the comparison to our test model. Finally, the fatigue testing is carried out with the established GRNN model. In conclusion, the GRNN model, whose recognition accuracy under the fatigue state has peaked at 88.1%, gets the better results than the SVM model. It has a better effect on the EEG fatigue detection in the sense of stability and discrimination.更多还原