基于行为和生理指标的驾驶风险分析与模式识别研究

发布时间：2018-07-03 11:24

本文选题：交通事故 + 人为因素　；参考：《清华大学》2015年博士论文

【摘要】：交通事故及其造成的人员伤亡是全球性的社会问题。驾驶员的人为风险因素不仅是造成交通事故的内在关键原因,而且受到外在环境因素的影响。为深入分析内外风险因素相互作用的事故致因机理,本论文在风险平衡理论的框架下设计了一系列离线和在线实验,综合采用行为与生理指标测量驾驶员在具体环境中的“感知-认知-反应”过程,并在此基础上采用模式识别方案预测交通事故的发生。在离线任务方面,本文基于实路视频的风险驾驶行为研究表明:驾驶员间的道路资源竞争导致普遍的风险驾驶行为,而风险驾驶行为既是驾驶员本人的内在个体风险,又是他人的外在环境风险。本文基于离线风险感知的评估结果表明:风险驾驶行为与风险感知水平呈现出负相关关系;而离线风险决策任务和反馈事件相关电位指标的研究结果表明:高风险行为倾向驾驶员对高不确定性和冒险性的选项具有较高的偏好,且其认知决策阶段的脑功能反应表现出了较高的正反馈激励和较低的负反馈抑制。以上离线研究的结果表明,针对驾驶员的感知和认知特点进行干预,可以有效降低其驾驶行为的风险程度,进而降低交通事故机率。而在线任务方面,本文模拟驾驶任务的研究结果表明:驾驶员根据道路环境自发调节其行为,以保持内外风险的平衡状态。本文开展了三类情景下的在线研究:在无风险情景中,驾驶员自由选择其行为;在风险决策情景中,驾驶员作出抢行或让行的决策;在险情情景中,驾驶员进行风险规避。高风险行为倾向驾驶员表现出了较多的抢行决策和区别于其他驾驶员的行为模式;发生事故驾驶员表现出较低的风险预警能力,其导致的生理反应可由皮电上升率指示。以上在线研究的结果,揭示了不同道路环境中内外风险失衡的原因和事故致因机理。在模式识别研究方面,本文采用不同特征组合和不同判别分析模型,验证了采用行为与生理指标实时预测交通事故的可行性。结果表明,包含所选行为与生理指标特征组合的模式识别方案具有较高的准确率和较低的误报率,在此基础上开发的事故预测系统可以综合考虑内外风险因素,准确地预测事故的发生。本论文研究综合采用行为与生理指标测量驾驶过程中的风险因素,揭示了不同环境下驾驶行为背后的深层次心理与生理机理,扩展了现有的风险平衡理论框架。研究提出的事故模式识别方案,可以直接应用于实时在线的交通事故预测,对其他人机系统状态识别研究亦有借鉴作用。
[Abstract]:Traffic accidents and the casualties caused by them are a global social problem. Driver's human risk factor is not only the internal key cause of traffic accident, but also affected by external environmental factors. In order to deeply analyze the mechanism of the interaction of internal and external risk factors, a series of off-line and on-line experiments are designed in this paper under the framework of risk balance theory. The behavioral and physiological indexes were used to measure the process of perception, cognition and response of drivers in specific environment, and the pattern recognition scheme was used to predict the occurrence of traffic accidents. In the field of off-line task, the research of risk driving behavior based on real road video shows that the competition of road resources among drivers leads to universal risk driving behavior, and the risk driving behavior is not only the inherent individual risk of the driver himself, but also the risk driving behavior. It is also the external environmental risk of others. The evaluation results based on off-line risk perception show that: the risk driving behavior is negatively related to the risk perception level; The results of off-line risk decision-making tasks and feedback event-related potential indicators show that high-risk behaviors tend to have a higher preference for high-uncertainty and risk-taking options. And its cognitive decision-making stage of brain function response showed a higher positive feedback incentive and lower negative feedback inhibition. The results of the above off-line studies show that intervention based on the perception and cognitive characteristics of drivers can effectively reduce the risk of driving behavior and then reduce the probability of traffic accidents. In the field of online task, the result of simulating driving task shows that the driver adjusts his behavior spontaneously according to the road environment in order to keep the balance of internal and external risks. In this paper, three kinds of online research are carried out: in the risk-free scenario, the driver freely chooses his behavior; in the risk-decision-making scenario, the driver makes the decision of robbing or letting the driver move; in the case of risk, the driver makes the risk aversion. The drivers with high risk behavior tend to show more robberies and behavior patterns different from other drivers, and drivers with accidents show lower risk early warning ability, and the physiological responses can be indicated by the increase rate of skin electricity. The results of this online study reveal the causes of the imbalance of internal and external risks and the mechanism of accidents in different road environments. In the aspect of pattern recognition, different feature combinations and different discriminant analysis models are used to verify the feasibility of real-time prediction of traffic accidents with behavioral and physiological indicators. The results show that the pattern recognition scheme, which includes the combination of selected behavior and physiological indicators, has higher accuracy and lower false alarm rate, and the accident prediction system developed on this basis can comprehensively consider internal and external risk factors. Accurately predict the occurrence of an accident. In this paper, the risk factors in driving process are measured comprehensively by using behavioral and physiological indexes, which reveals the deep psychological and physiological mechanism behind driving behavior in different environments, and extends the existing theoretical framework of risk balance. The proposed scheme of accident pattern recognition can be directly applied to real-time and on-line traffic accident prediction, and can also be used as a reference for other man-machine system state recognition research.
【学位授予单位】：清华大学
【学位级别】：博士
【学位授予年份】：2015
【分类号】：U492.8

【相似文献】