论文部分内容阅读
为有效识别驾驶员疲劳状态,基于脑电信号(electroencephalogram,EEG)提出了一种驾驶疲劳状态识别方法.首先,以时间段划分疲劳等级,并采用主、客观测评指标对疲劳等级划分的合理性予以验证.然后,利用快速傅里叶变换对脑电信号进行分析,在此基础上选取3种频段的平均幅值和5项合成指标,通过核主元分析(kernelprincipal component analysis,KPCA)构建疲劳识别脑电指标,结合支持向量机(support vector machine,SVM),构建了驾驶员疲劳状态识别模型.最后,采用30名驾驶员连续驾驶2 h的脑电数据,对该模型方法进行试算.试算结果表明:疲劳状态识别正确率在79.17%~92.03%,平均正确率为84.62%,该方法可用于驾驶疲劳识别.
In order to effectively identify the driver’s fatigue status, a driving fatigue status recognition method based on electroencephalogram (EEG) is proposed.Firstly, the fatigue level is divided by time period, and the rationality of the classification of fatigue level by using the objective and objective evaluation indexes Then, the EEG signals were analyzed by using Fast Fourier Transform, and then the average amplitudes and five composite indexes of the three frequency bands were selected and the fatigue was built by kernel principal component analysis (KPCA) The EEG index was identified and a driver fatigue status recognition model was constructed based on support vector machine (SVM) .Finally, 30 drivers were continuously piloted for 2 hours to test the model method. The test results show that the correct rate of fatigue state identification is 79.17% ~ 92.03%, and the average correct rate is 84.62%. This method can be used for driving fatigue identification.