基于组合近似模型的商用车驾驶室正碰安全性优化

Safety Optimization of Commercial Vehicle Cab Frontal Collision Based on Ensemble of Surrogate Models

相较于乘用车,商用车重量大、碰撞能量大,驾驶室前部缺少足够的吸能区,乘员生存空间极易受到挤压,正碰安全性优化十分必要。在现有的研究中,往往通过大量采点构造单一近似模型,来优化驾驶室碰撞安全性。提出将极限学习机模型与响应面、径向基神经网络、支持向量回归模型组合,并引入遗传算法匹配4个模型的权系数。采用上述组合近似模型,结合多目标优化算法,以生存空间尺寸和驾驶室质量为优化目标,在厚度和材料的混合变量空间下寻优。结果表明:组合近似模型精度整体上高于单一近似模型,优化后驾驶室的正碰生存空间显著增大,在正碰、顶压与后推3种工况下的安全性均符合国家标准的要求,同时总体减重7.68kg。

Compared with passenger cars,commercial vehicles have large weight and collision energy,and the front part of the cab lacks sufficient energy absorption area,so the living space of occupants is easily squeezed.Therefore,it is very necessary to optimize the forward collision safety.In the existing studies,the single surrogate model is usually constructed by collecting a large number of points to optimize the cab crash safety.The extreme learning machine model is combined with response surface model,radial basis function neural network model and support vector regression model,and the weight coefficients of the four models are matched by genetic algorithm.The ensemble of surrogate models and the multiobjective optimization algorithm are adopted,and the size of the living space and the mass of the cab are taken as the optimization objectives,and the optimization is carried out in the mixed variable space of thickness and material.The results show that the overall accuracy of the ensemble of surrogate models is higher than that of the single surrogate model.After optimization,the forward collision living space of the cab is significantly increased,and the safety under the three working conditions of forward collision,top pressing and backward pushing all meet the requirements of national standards.Meanwhile,the overall weight of the cab is reduced by 7.68 kg.