【作者】 施颐；

【导师】 朱平；

【作者基本信息】 上海交通大学 ， 车辆工程， 2010， 硕士

【摘要】 节约资源和减少环境污染是世界汽车工业界亟待解决的两大问题,汽车轻量化是实现节能和环保的有效途径。车身重量约占整车重量的40%左右,因此车身轻量化对于整车的轻量化起着举足轻重的作用。车身轻量化设计是多参数、多约束的复杂系统优化问题,首先需要满足耐撞安全性。借助传统有限元分析和“试错法”等手段进行优化设计容易造成计算量过大、优化收敛缓慢甚至无法得到全局最优解等缺点,应用近似模型代替有限元仿真来预测或拟合结构性能响应是国内外研究的前沿和热点。近似模型是一种能够反映设计变量与性能响应之间函数关系的数学模型,通常是结合试验设计和近似拟合方法建立的。目前,根据不同领域的工程应用,对试验设计和近似拟合方法的选择上具有一定的随意性和不确定性,因此车身结构轻量化设计需要对相应的建模方法进行合理评价和选择。本文针对现有车身结构形式,利用近似模型理论结合数值优化算法,对车身结构轻量化设计进行了较为系统的研究,在确保刚度及耐撞性能的前提下,实现车身结构的轻量化。主要研究工作及结论如下:1)针对车身结构耐撞性能响应具有强非线性的特点,根据近似模型的精度检验指标——均方根误差和最大绝对误差,对比分析均匀设计、拉丁超立方设计及汉默斯里序列抽样这三种常用的现代试验设计方法的抽样方式对支持向量回归近似拟合方法建模的影响。通过对比研究,在满足轻量化设计的“小样本规模”范围条件下,将均匀设计和支持向量回归方法相结合,最适合建立车身结构耐撞性能指标的近似模型。2)在经典优化理论的基础上,提出了基于近似模型的车身结构轻量化设计流程:结合均匀设计和支持向量回归方法能够建立具有良好整体预测精度的近似模型;基于自适应过程的优化设计通过提升近似模型在最优解处的预测精度,确保得到真实的全局最优解;针对轻量化最优解进行合理的工程修正及仿真验证,能够确保设计方案的工程可行性。根据该设计流程,以车身关键部件——车门为例进行轻量化设计,在确保车门结构刚度及耐撞性能的前提下,实现6.72%的轻量化效果,验证了轻量化设计流程的可行性。3)针对整车车身结构复杂系统具有性能类型多样且各性能响应非线性程度不一的特点,以某型轿车为例,综合考虑弯曲刚度及侧碰耐撞性能,利用均匀设计和支持向量回归方法建立结构性能指标的近似模型,通过遗传算法优化车身零件的厚度及相应的材料参数,实现车身结构轻量化设计,减少的质量为9.1kg,实现了5.44%的轻量化效果。更多还原

【Abstract】 Energy saving and environmental protection are two main problems for the world automobile industry and vehicle lightweight is an effective way to realize these two problems. Weight reduction of body structure plays a rather important role in lightweight of full vehicle, for body structure weights about 40 percent of full vehicle. Autobody lightweight design is an optimization procedure with multiple parameters and constraints which should firstly satisfy crashworthiness performance. Traditional design via finite element analysis and“trial and error”method has the disadvantages of huge computational cost and slow convergence, then the optimization design in which approximation model is employed in lieu of finite element simulation becomes a crucial and hot research topic. The approximation model which is built by experimental design and approximation method can express the relationship between design variables and responses. Nowadays the selection of these methods which are used to build approximation model is to some extent arbitrary and uncertain, so this problem should be considered properly for the autobody lightweight design. This dissertation presents a systemic study on autobody lightweight design based on theories of approximation model and optimization method. The purpose of the lightweight design is to achieve weight reduction of the original body structure meanwhile meeting both stiffness and crashworthiness requirements. The studying work is summarized as follows:1) In the view of high nonlinear characteristics of crashworthiness responses, the comparison of three types of experimental design methods in terms of their capability to generate accurate approximation models for crashworthiness indicators is conducted based on Support Vector Regression method. The result reveals that uniform design integrated with Support Vector Regression method can provide the most accurate approximation model while the number of sample points is limited into the“small scale range”which is suitable for autobody lightweight design.2) A feasible flow for autobody lightweight design based on approximation model is proposed. In the design flow, approximation model which is built by uniform design and Support Vector Regression method will ensure its global accuracy. The adaptive optimization procedure will improve the local accuracy of approximation models around the optimal result. The proper engineering modification will ensure the feasibility of lightweight result. According to the design flow, a lightweight design for car door structure is conducted which not only realizes the weight reduction effect of 6.72%, but also proves the feasibility of the design flow.3) The full vehicle structure is a complex system and it contains various performances which have different levels of nonlinearity. The lightweight design for full vehicle body structure considering both stiffness and crashworthiness is conducted by optimizing the sheet thickness and the corresponding material parameters of body components. As a result, the reduced weight of body structure is 9.1kg which realizes the weight reduction effect of 5.44%.更多还原