基于EEG频谱特征的驾驶员疲劳监测研究

研究表明疲劳驾驶是引发交通伤亡事故的重要原因之一,因此有必要采取相应的预防措施。脑电是公认的睡眠(疲劳)金指标,因此论文提出了基于脑电频谱特征的驾驶员疲劳预测方法。采用了驾驶模拟实验中记录的三路驾驶员脑电信号,并利用驾驶员自评与专家评定两种方式相结合的方法将驾驶数据分为疲劳和清醒。针对脑电中眼电噪声很强的特点,对记录的脑电进行了自适应滤波消噪处理,结果显示可有效滤除眼电伪迹;然后根据脑电的频域特征比较突出且与疲劳相关的特点,从去噪后的脑电中提取出了的75个频谱特征;最后利用这些频谱特征,采用朴素贝叶斯分类的方法建立了驾驶员疲劳监测模型。实验结果表明,该方法能监测出驾驶员84%的疲劳状态。

Study on driver fatigue detection based on EEG spectrum-related features

Various investigations showed that drivers＇ drowsiness is one of the main causes of traffic accidents.Thus,some countermeasure were currently required in many fields for sleepiness related accident prevention.In this paper the drowsiness prediction was performed with EEG spectrum parameters,which is believed to be the golden standard of sleep/fatigue detection.Firstly,EEG data of 3 channels was collected in a driving simulation experiment and the driver＇s drowsiness levels were labeled as fatigue and alert by self rating of the drivers and the expert assessment;secondly,adaptive filtering was carried out to remove EOG artifacts which were the dominate background noise of EEG recordings and then 75 EEG spectrum related features were extracted from the clean EEG.In the end,a driver fatigue detection model was established on the EEG features by Naive Bayes Classifier.The results showed that our approach can detect 84% of drivers＇ fatigue.