基于脑电信号的驾驶持续性注意水平识别方法研究

发布时间：2018-01-25 08:29

本文关键词： 驾驶持续性注意水平脑电信号支持向量机识别特征指标　出处：《西南交通大学》2017年硕士论文　论文类型：学位论文

【摘要】：驾驶持续性注意水平是影响驾驶安全的关键性因素,而脑电信号的变化与驾驶持续性注意水平存在着直接联系,因此为了对驾驶持续性注意水平予以有效识别,本文以脑电信号作为识别特征指标,构建了一种驾驶持续性注意水平识别方法。首先,为了对驾驶持续性注意水平进行等级划分,本文采用驾驶反应时间作为客观划分指标,提出了一种驾驶持续性注意水平等级划分方法。在此基础上,利用高效的筛选算法对脑电特征参数进行择优筛选,所得到的脑电特征参数作为驾驶持续性注意水平识别特征指标。最后,采用粒子群算法对支持向量机进行参数优化,构建了驾驶持续性注意水平识别模型。基于已有的模拟驾驶平台,设计了驾驶持续性注意水平实验,将实验所采集的脑电数据对该模型予以试算,得到如下结论:(1)采用反应时间对驾驶持续性注意水平进行等级划分,将从属不同等级的驾驶行为绩效进行差异性对比分析,分析结果认为不同等级的驾驶行为绩效存在着显著差异,验证了文中所提方法的合理性。(2)采用Kruskal-Wallis检验和Relief算法对脑电特征参数进行择优筛选,得出位于FP1电极的合成参数(α+θ)/β、P7电极的合成参数θ/β、CZ电极的合成参数(α+θ)/β、P8电极的合成参数(α+θ)/β及FZ电极的合成参数(α+θ)/β在不同驾驶持续性注意水平等级下差异最为显著,可作为驾驶持续性注意水平的识别特征指标。(3)模型识别平均正确率为92.19%,说明所提模型和方法的可行性和适用性,认为可用于对驾驶持续性注意水平的识别。
[Abstract]:Driving sustained attention level is the key factor to affect driving safety, and the change of EEG is directly related to driving sustained attention level, so in order to identify driving sustained attention level effectively. In this paper, the EEG signal is used as the recognition feature index, and a method of recognition of driving persistent attention level is constructed. Firstly, in order to classify the driving sustained attention level. In this paper, the driving reaction time is used as the objective classification index, and a method of classification of driving persistent attention level is proposed. On the basis of this, an efficient screening algorithm is used to select the characteristic parameters of EEG. The obtained EEG feature parameters are used as the recognition feature index of driving sustained attention level. Finally, the particle swarm optimization algorithm is used to optimize the parameters of support vector machine. Based on the existing simulated driving platform, a driving sustained attention level experiment was designed, and the EEG data collected from the experiment were used to test the model. The conclusion is as follows: 1) the reaction time is used to grade the driving sustainability attention level, and the performance of driving behavior belonging to different grades is compared and analyzed. The results show that there are significant differences in driving behavior performance among different grades. The rationality of the proposed method is verified. (2) Kruskal-Wallis test and Relief algorithm are used to select the characteristic parameters of EEG. The synthesis parameters of FP1 electrode (伪胃 O / 尾 P7 electrode) are obtained, and the synthesis parameters of CZ electrode (伪胃 n / 尾) are obtained. The synthesis parameters of P8 electrode (伪胃 O / 尾) and FZ electrode (伪胃 N / 尾) showed the most significant difference under different levels of continuous driving attention. The average recognition accuracy of the model is 92.19, which shows the feasibility and applicability of the proposed model and method. It is believed that it can be used to identify the level of continuous attention to driving.
【学位授予单位】：西南交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U491.25

【参考文献】