汽车车身基于大数据的结构-材料-性能一体化设计方法研究

Multidisciplinary optimization does not fully solve the key design problems of automobile body structure without consideration of material characteristics. Therefore, body structure and material of automobile, which could balance two conflicting objectives, such as lightweight and crashworthiness, under uncertainty in the process of design and manufacture, is researched in this topic. A theoretical method of design optimization of structural-material-performance coupling based on Big Date is to be proposed. In the process of building and iterative optimization the mathematical model, it mainly studies the real-time design point and distribution trend of characteristic values of uncertainty coupling variables based on fitting of all sample data in big-set, and the accurately description method for uncertainty variables based on the characteristic values of those variables, and the mechanism of variable coefficient consistency constraints reacted on shared coupling variables which belongs to multiple disciplines for the concurrent optimization design, and the sequential dynamic search optimization method using the real-time design point of uncertainty variable to make the uncertainty multidisciplinary coupling design optimization strategy. It will be realized the reliability and robust design of the body structure, and then be obtained optimization solution for the comprehensive properties (crash safety, driving stability and lightweight) of automobile. The research will shorten the development cycle of automobile products, and enhance the competitiveness of products. The combination of Big Data and vehicle design will has great theoretical significance and engineering practical value in the design and related field of automobile products.

缺乏材料因素的多学科优化并未充分地解决汽车车身结构关键设计问题，为此，本项目协同车身结构与材料，以平衡车身结构轻量化与耐撞性之间的冲突为优化目标，分析设计、生产装配过程中车身结构、材料、性能参数受不确定性因素的影响，提出一种基于大数据的结构-材料-性能一体化设计方法，通过数学模型建立与迭代寻优，重点研究拟合实时更新的不确定性耦合变量大数据信息来获取变量特征参数的分布趋势和时变设计点位置，建立基于特征参数分布的不确定性变量精确描述方法，探讨协同优化时不同学科之间耦合变量的变系数系统一致性约束机理，探究基于不确定性变量时变设计点位置的序列动态搜索寻优方法，制定不确定性多学科耦合设计优化策略，基于可靠性与稳健性设计，实现车身轻量化与耐撞性综合性能最优方案，缩短产品开发周期，提升产品竞争力，拓展车辆现代设计进入大数据时代，在汽车产品设计开发及相关领域具有重大的理论意义和工程实用价值。

开发汽车桁架车身系统基于大数据的结构-材料-性能一体化设计方法，对综合提高整车轻量化、碰撞安全性、NVH及气动特性有重要意义。本项目针对FSEC赛车桁架车身系统进行基于大数据的不确定性多学科优化设计框架研究，主要研究内容有：通过机器学习对系统设计变量进行大数据预处理，根据训练集数据验证测试集数据模型精度，采用高斯随机过程构建近似模型，增加超参数惩罚项进行模型补偿，有效提高近似模型精度；建立基于动态松弛因子的目标分流法，根据迭代过程中层级学科间耦合变量不一致信息自适应调整动态松弛因子，保证设计空间的可行域，改善目标分流法的收敛性；完成FSEC赛车在加速、制动、侧向加速等工况的强度分析，采用等效静载荷法分析赛车桁架结构碰撞过程传力路径并对车身结构优化改进，采用挠跨比对单一材料车身杆件变形进行评价，采用冒泡法编制计算机可执行的桁架结构正向选材策略，将单一材料车身替换为多材料车身，实现整车耐撞性与轻量化综合提升；提出FSEC赛车车身结构-材料-性能一体化设计框架，建立高精度近似模型、动态系统一致性约束、与之相匹配的多学科耦合设计优化策略，采用6σ理论进行稳健性优化设计，为实现桁架式车身系统综合性能匹配与快速设计提供了新的思路和方法。