将概率模型与基于相关性分析技术建立的凸模型有效结合一起,并应用到汽车碰撞安全可靠性分析中。将汽车碰撞中主要吸能部件的几何尺寸、材料参数作为不确定变量,根据其样本信息获取难易程度,分别使用概率模型和椭球凸模型来描述。应用最优拉丁超立方采样方法在不确定区间内进行采样,通过二次多项式响应面法拟合碰撞过程中各响应的近似模型,基于混合模型计算汽车碰撞安全性的可靠度指标。与参数确定性分析相比,基于不确定变量混合模型的可靠性分析更真实地反映客观实际,为提高汽车安全性设计提供依据和指导。
The convex model based on correlation analysis technique is applied to the automobile collision safety reliability analysis combined with the probability model. Make the key component's structure sizes and material parameters in the vehicle collision as the uncertain variable, according to its sample information acquisition with difficulty degree use probability model and convex model describe them. The Latin Hypercube sampling method was adopted to generate the sampling points, and then a second-order respond surface was constructed to approximate the vehicle collision response. The reliability index of the vehicle collision safety based on the hybrid model was obtained. Comparing with the determinate analysis, the reliability-based hybrid model analysis can better reflect the real situation, and provide an instruction for the vehicle safety design.