中文摘要：

对22名动车组司机进行4h的模拟驾驶试验,采集参试司机完成动车驾驶及对随机信号的反应2项试验任务过程中在16～46和210～240min这2个时段内的主观疲劳测评数据、行为绩效数据和脑电数据,验证了将这2个时段分别作为动车组司机高、低持续性注意水平时段的合理性;采用快速傅里叶变换方法分别提取4～8,8～13和13～30Hz这3个频段脑电数据的平均幅值,进行脑电熵的计算;采用Kruskal-Wallis检验和Relief算法,选取差异性最显著或权重最大的脑电熵作为动车组司机持续性注意水平的敏感性指标。研究结果表明：与第1时段相比,动车组司机在第2时段（210～240min）的疲劳程度显著增加,对列车运行速度的控制能力和对随机信号的反应能力显著下降;在贴于头皮上的FP1和F7电极处频段为13～30 Hz的香农熵以及FZ电极处频段为8～13Hz的样本熵对动车组司机持续性注意水平的影响十分敏感。

英文摘要：

22EMU drivers attended to driving simulation experiment for 4h,the subjective fatigue evaluation data,behavior performance data and electroencephalogram（EEG）data were collected in two time intervals of 16～46 and 210～240min when 22 EMU drivers completed the two experimental tests,including a driving task and a random signal task.The rationality was verified that the two time intervals were respectively used as the time intervals of high and low sustained attention levels.Fast Fourier Transform（FFT）was adopted to respectively extract the mean amplitude of EEG data in three bands of 4～8,8～13 and 13～30 Hz to calculate the value of EEG entropy.Kruskal-Wallis test and Relief algorithm were adopted to select EEG entropy with the most significant difference or the maximum weight as the sensitive index for the sustained attention level of EMU drivers.The results showed that the fatigue level of EMU drivers in the second time interval（210～240min）was significantly increased than that in the first time interval（16～46min）,and the control ability for train running speed and the reaction ability to random signal were significantly decreased.The EEG Shannon entropy of FP1 and F7 electrode affixed to scalp in the bands of 13～30Hz,and the EEG sample entropy of FZ electrode in the band of 8～13Hz were very sensitive to the sustained attention level of EMU drivers.