基于可穿戴设备与脑电信号的高速轨道交通车辆驾驶员疲劳预警研究

The driver's vigilance plays an important role in traffic safety. Eectroencephalogram(EEG) signal is the important technique in driver vigilance study. With the development of high-speed train in the world, Brain electric activity research for driver vigilance gets more and more attention. But in realistic driving conditions, the problems of driver fatigue detection based on EEG limit it’s applications because of invasive operation, non real-time and reliability. This project will study four key technologies of this project will be combined with the actual needs of high-speed rail vehicle safety operation. The first research on innovation of train vigilance detection is the wireless EEG signals are collected and Compressed sensing de-noising processing. The second key research is high-speed train driver’s brain signal models and vigilance describes based on electroneurophysiology. The third innovation is the fatigue factor classifying method based on fatigue factor classification algorithm which used discriminantive dictionary and weighted local features analysis method((2D)2PCA). The last research is the high-speed train driver vigilance detection technique based on multi-biometric recognition. This project aims to detect and prevent the super high-speed train driver fatigue driving. And this system can provide the theoretical support and technical basis to the vigilance detection equipment of brain computer interface.

驾驶员的警觉度在交通安全中起着重要的作用，脑电信号是评估驾驶员警觉度状态的重要指标，随着对高速轨道交通车辆驾驶员自身状态关注度的提高，基于脑电信号的驾驶员警觉度研究越来越受到重视。但在现实驾驶环境下，基于脑电信号警觉度预测技术的侵入性、非实时性和可靠性问题限制了其应用。本项目将结合高速轨道交通车辆安全运营的实际需求，应用无线可穿戴EEG技术采集驾驶员脑电信号并进行去噪预处理；应用神经电生理理论，建立驾驶员脑电信号数学模型，并对驾驶员的警觉度进行标注；采用双向二维子空间方法进行疲劳分类，提出二维加权局部保持判别式分析方法(2D)2PCA的疲劳因子分类算法；探索融合多模态生物特征的驾驶员警觉度预测及告警技术。其研究成果为检测和预防高速轨道车辆驾驶员疲劳驾驶、基于脑电信号的警觉度预警设备研制和应用提供理论基础和技术支持。

驾驶员的警觉度在交通安全中起着重要的作用，脑电信号是评估驾驶员警觉度状态的重要指标，随着对高速轨道交通车辆驾驶员自身状态关注度的提高，基于脑电信号的驾驶员警觉度研究越来越受到重视。但在现实驾驶环境下，基于脑电信号警觉度预测技术的侵入性、非实时性和可靠性问题限制了其应用。.因此，本项目结合高速轨道交通车辆安全运营的实际需求，围绕驾驶员脑电信号采集和疲劳驾驶分析预测展开研究。主要创新包括：（1）应用无线可穿戴EEG技术采集驾驶员脑电信号并进行去噪预处理；（2）应用神经电生理理论，建立驾驶员脑电信号数学模型，并对驾驶员的警觉度进行标注；（3）采用双向二维子空间方法进行疲劳分类，提出疲劳因子分类算法；（4）探索融合多模态生物特征的驾驶员警觉度预测及告警技术。其研究成果为检测和预防高速轨道车辆驾驶员疲劳驾驶、基于脑电信号的警觉度预警设备研制和应用提供理论基础和技术支持。