基于多源生理信号的驾驶疲劳检测

Driving Fatigue Detection Based on Multi-source Physiological Signal

为检测驾驶疲劳,基于移动穿戴设备采集了4h模拟驾驶的生理信号(肌电、皮电、心率、血氧饱和度),分析各生理信号确定疲劳阈值,然后将其融合建立驾驶疲劳检测神经网络模型,依据被试者对刺激信号做出反应的时间,并通过脑电指标δ/β和δ对模型进行检验。结果表明,在长时间模拟驾驶过程中,疲劳是一种突变行为;各生理信号可反映驾驶疲劳;多源生理信号融合提高了驾驶疲劳检测模型的准确率,可用于开发可穿戴设备。

In this paper, for the purpose of driving fatigue detection, a 4-hour simulated driving test was carried out, in which the physiological signals, such as the electromyography, skin electricity, heart rate and oxygen saturation signals of the subjects, were collected using wearable devices. Then the fatigue thresholds were determined by analyzing physiological signals and were incorporated to establish the driving fatigue testing neural network model, based on which, the EEG indicator δ/β and δ-pair model of the subjects were tested according to the necessary time for them to respond to the stimulus signal. The result showed that during long-time simulated driving, fatigue is a type of abrupt behavior; the physiological signal can reflect driving fatigue; and the incorporation of the multi-source physiological signals can improve the accuracy of the driving fatigue detection model and be used in wearable device development.