摘要
利用生理信号的无线测量设备实现了对驾驶员在驾驶过程中的脑电信号、肌电信号和呼吸信号的采集,并对其进行分析处理,从而实现驾驶员的疲劳检测.首先分别计算三个生理信号的近似熵并将其作为疲劳检测的特征参数,然后使用主成分分析对特征参数进行降维优化处理,同时对原始特征参数和分析后的主成分分别进行统计分析,基于优化处理后的特征参数利用回归方程建立驾驶疲劳估计模型.最后通过交叉验证对本方法进行评价,并使用数据融合方法给出了综合的评价结果.评价结果表明提出的方法对驾驶员疲劳状态的检测正确率达到90%以上.
Driver fatigue was detected using electroencephalograph, electromyography and respiration signals, which were collected wirelessly. The approximate entropies of the three signals were selected as features, and the reduction of feature dimensions was achieved by principle component analysis. Statistical analyses were then given to both original features and principle components and an evaluation model for driver fatigue was established using regression equation. The experimental results were evaluated by cross validation and the accuracy was more than 90% based on data fusion method. The results verify that the model is effective in detecting driver fatigue.
出处
《东北大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2014年第6期850-853,共4页
Journal of Northeastern University(Natural Science)
基金
国家自然科学基金资助项目(61071057)
关键词
疲劳驾驶
无线体域网
脑电
肌电
呼吸
driver fatigue
wireless body area network
electroencephalograph
electromyography
respiration
