A research on vehicle accident reconstruction using inverse problem analysis

汽车碰撞事故是当今社会的严重问题。事故重建包含从碰撞后车辆静止位置寻找车辆PM碰撞速度的过程。这被理解为从事后结果重建事件原因的逆问题之一。本研究的目的是建立汽车碰撞事故的反分析方法，将人工神经网络技术应用于汽车碰撞事故的反分析。本研究的结果如下：1。将车辆碰撞事故分为两个阶段：车辆碰撞和碰撞后运动。利用人工神经网络对每一阶段进行了反分析。通过顺序使用这两个逆分析，从最终的车辆静止位置重建了碰撞前的车辆速度。在本研究中，车辆的碰撞前速度被重构为…对直角侧面碰撞事故的合理精度要求更高。这一次，只审查了直角碰撞事故。但是，这种方法的有效性必须在近夹具的其他撞击构型上进行检验。在人工神经网络的学习过程中，利用CarSim仿真软件建立了详细的车辆动力学模型，生成了碰撞后车辆运动的监督信号。另一方面，使用刚体碰撞理论来生成车辆碰撞速度的监督信号。刚体碰撞理论可以给出碰撞前和碰撞前速度的物理合理但实际粗略的估计。因此，这一次的研究结果对于实际重建来说并不具有足够的准确性。但是，它们可以作为目前详细重建的第一个估计。当有了更准确、更系统的监测信号时，这种方法的精度自然会提高。当然，本研究所构建的框架仍然适用于此类CAWS，并有可能更少地进行更准确的反分析

Accidents of vehicle collision are serious problem in our society today. Accident reconstruction contains the process that seeks the vehicle pm-impact speeds from the post-impact vehicle rest position. This is understood as one of the inverse problems which reconstruct the cause of the incident from the post-incident results. The purpose of this research is to construct the method of inverse analysis for the vehicle collision accidents, lb solve the vehicle collision accident inversely the artificial neural network technique was used in this research. The results of this research are as followings1. The vehicle collision accident was divided into two stages: vehicle collision and post-impact motion. The inverse analyses were carried out by using artificial neural network for each stage. By using the both inverse analyses sequentially, the pre-impact vehicle speeds were reconstructed from the final vehicle rest positions. In this research, the vehicle pre-impact speeds were reconstructe … More d in the reasonable accuracy for right angled side impact accident. This time, only the right angled collision accidents were examined. But the validity of this method has to be examined for other impact configurations in near fixture.2. In the learning process of the artificial neural network, supervised signals for the post-impact vehicle motion were made by detailed vehicle dynamics model using CarSim simulation software. On the other hand, the rigid body collision theory, which may give physically rational but practically rough estimations of pre- and poet- impact speeds, was used to make the supervised signal for the vehicle impact speeds. Therefore, the results of this time of research do not have enough accuracy for practical reconstruction. However, they may be used as a first estimation for the detailed reconstruction at the present.3. When more accurate and systematic supervised signals axe available, the accuracy of this method will increase naturally. Of course, the frame work which constructed in this research will be still valid for such caws and have a possibility to carry out more accurate inverse analyses Less