论文部分内容阅读
平乏、单调的监控作业容易使作业人员觉醒水平下降,为提高监控工作的绩效,需识别及唤醒低觉醒状态,因此本文提出以脑电信号(EEG)为研究对象的低觉醒状态识别方法。运用小波包变换分解警戒作业人员的EEG信号,获取EEG信号中的δ、θ、α、β等节律成分;结合各节律计算相对能量和高低频能量比参数等特征,组成低觉醒状态识别的特征向量,并使用支持向量机对模拟警戒作业中的低觉醒状态进行了识别。实验结果显示,本文方法能够很好地区分警戒作业中的低觉醒状态和觉醒状态,识别率高。与其它分析方法相比,该方法能够有效地识别警戒作业中的低觉醒状态,能够为低觉醒状态的唤醒机制提供技术支持。
In order to improve the performance of monitoring work, it is necessary to identify and awaken the low awakening state. Therefore, this paper proposes a low awakening state recognition method based on EEG (EEG). Using wavelet packet transform to decompose the EEG signals of alerting workers to obtain the rhythmic components of δ, θ, α, β in EEG signals; combining the rhythms to calculate the characteristics of relative energy and high and low frequency energy ratio parameters to make up the features of low arousal state recognition Vector, and uses support vector machines to identify the low wakefulness in simulated alerting operations. Experimental results show that this method can distinguish between low awakening status and awakening status in alerting operation with high recognition rate. Compared with other analysis methods, this method can effectively identify the low awake state of alerting and provide technical support for the wake-up mechanism of low awakening.