一种分析事故再现仿真结果不确定性的多响应面-均匀设计法

发布时间：2018-07-02 15:13

本文选题：车辆安全 + 事故再现　；参考：《汽车工程》2017年01期

【摘要】：交通事故现场的痕迹由于受到其他车辆和行人等外界环境的影响而产生不确定性,但若将这些不确定性信息反映到事故再现结果中,即可增强再现结果的可信性。为能更好地从不确定痕迹定义域空间内找到事故再现仿真结果的取值区间,在均匀设计的基础上,提出一种改进的多响应面-均匀设计法(MUD法)。该方法首先用均匀设计生成实验样本点并进行实验;然后借助正交设计中的极差分析法分析实验结果而找出可能产生极值的子空间域;再在子空间域内生成新的样本点,并依托实验结果分析各子空间域及整个定义域空间内事故再现结果的极值;最终综合这些极值给出再现结果之取值区间。通过一个数值算例及真实的车人碰撞事故案例,发现MUD法能显著改善事故再现结果的精度,且仅需较少的仿真次数即能找出可能产生极值的子空间域。
[Abstract]:The trace of the scene of a traffic accident is uncertain due to the influence of other vehicles and pedestrians, but if the uncertainty information is reflected in the result of the accident reappearance, the credibility of the reappearance result can be enhanced. The value of the simulation result of the accident reappearance can be found in the space of the uncertain trace definition domain. The interval, on the basis of uniform design, proposes an improved multi response surface uniform design (MUD method). The method first generates experimental sample points and experiments with uniform design, and then analyses the experimental results with the aid of the extreme analysis in the orthogonal design to find the subspace domain which may produce the extreme value; and then generates a new one in the subspace domain. The sample points and the experimental results are used to analyze the extreme values of the accident reappearance results in each subspace domain and the whole domain space. Finally, these extremes are integrated to give the interval of the value of the reappearance results. Through a numerical example and a real car crash case, it is found that the MUD method can significantly improve the accuracy of the accident reconstruction results, and only need less. The number of simulations can identify subspace domains that may produce extreme values.
【作者单位】：长沙理工大学汽车与机械工程学院;工程车辆安全性设计与可靠性技术湖南省重点实验室;公安部交通管理科学研究所;
【基金】：国家自然科学基金(51208065) 湖南省科技计划项目(2015JC3056) 工程车辆安全性设计与可靠性技术湖南省重点实验室开放基金(KF1506)资助
【分类号】：U491.31

【参考文献】