用于智能驾驶系统评价的乘员损伤模型

只有较少的交通事故数据资源被用于建立基于碰撞速度信息的乘员损伤模型，致使所得到的模型精度差。为此，提出了基于车辆变形深度的乘员损伤模型。对美国不同制造年代和车辆级别的事故数据进行聚类分析，论证出车辆变形深度与乘员损伤风险具有相关性。以车辆变形深度为自变量，通过回归分析得到乘员损伤模型。不同种类车辆的乘员损伤模型拟合精度R²约为0.9，证明了该模型的正确性。为进一步验证，以此模型为基础，评价智能驾驶系统的有效性。以自动紧急制动系统为例，对比基于变形深度和速度变化量信息2种方法的有效性计算结果。结果表明：2组结果的平均误差不超过1%，验证了基于变形深度的乘员损伤模型的准确性。该模型仅需要事故数据库中准确的变形深度信息，能够获得更多的事故数据支持，从而可以更好地适应于不同类别智能驾驶系统的评价需求。

Occupant Injury Model for Evaluating Intelligent Driving Systems

The occupant injury model is often characterized as a function of the velocity information in an accident. Only a small amount of the total traffic accident data resources is used to fit the occupant injury model, which results in the poor accuracy of the obtained model. An occupant injury model based on the vehicle deformation depth is presented in this paper. A cluster analysis of vehicles of different ages and grades is carried out with accident data from the US to demonstrate the interdependency between the vehicle deformation depth and the occupant injury risk. Considering the vehicle deformation depth as an independent variable, the occupant injury model is obtained by the regression analysis. The accuracy R² of the fitted injury models for different types of vehicles is about 0.9, which indicates the reliability of the model. This paper puts forward a safety benefit evaluation method of the intelligent driving system based on the injury model. An automatic emergency braking system is evaluated using the deformation depth and velocity information methods. The difference between the two results is below 1%, which verifies the accuracy of the injury model. The proposed model needs accurate deformation depth information in the accident database. Further, it can get more accident data support and can be better adapted to evaluation needs of various intelligent driving systems.