概念设计阶段车身梁类关键截面反演仿生算法及在电动汽车轻量化上的应用

The key corss-sections of the body have a decisive impact on lightweight, cost and the performance of the automobile. Because of less information in conceptual design stage, the design of cross-section is mostly based on experience. But it is often discovered that it does not meet the performance and lightweight requirements, therefore this " blind cross-section design-verification-repeatedly modified" positive trial-and-error design does not meet the needs of modern body design and lightweight. In response to these problems, under the background of the concept design phase, for a particular body section complex constraint conditions and design objectives, we establishe the body section inversion theory, study on the mechanism of the inversion of the mechanical properties of the cross section to the geometry information, simulate ant colony foraging behavior and set up the corresponding inversion algorithm, and fulfill "the optimization of cross-section-the inversion generated of geometry-verification- slightly modified" design. It is suitable to the car concept design stage. This method based on the size constraints of the body styling, interior space and general arrangement, and taking into account the molding and other aspects of the shape constraints. Its purpose is to take advantage of the finite point thought to fulfill the discretization of the feasible domain of cross-sectional design. The optimization problem in continuous space is transformed into combinatorial optimization problem. And use the control node to produce a lightweight cross-sectional shape which meet the performance requirements. The application of this method in lightweight of electric vehicles will lightweight the body and greatly offset the increase in battery weight, and reduces the dependence of the vehicle battery. It also reduces the cost of the vehicle, and will actively promote the popularity of the development of electric vehicles.

车身关键截面对汽车性能、轻量化和成本具有决定性的影响。概念阶段设计信息较少, 截面大多根据经验设计，盲目性很大，这种“截面盲目设计-验证-反复修改”的正向试错设计方法已经不能满足现代车身设计和轻量化要求。针对上述问题，在概念设计阶段，提出一种截面设计反问题，针对特定的车身截面复杂约束条件和轻量化目标，构建车身截面力学特性向几何信息的反演理论，开展反演机理的研究，仿生蚁群觅食的自然行为，建立相应的仿生算法，实现截面“力学性能优化-几何反演生成-验证-少量修改”的逆向定量设计。该方法引入车身造型、内部空间及总布置等尺寸约束，同时考虑截面零件制造等形状约束要求，利用有限点思想将截面设计的可行域离散化，将连续空间的寻优问题转化为组合优化问题,通过节点控制生成截面形状。通过本方法在电动汽车轻量化上的应用，弥补电池所带来的重量增加，减少整车对电池的依赖程度，降低整车成本,推动电动汽车发展普及。

汽车概念设计阶段是车身设计的关键环节，尤其是车身中的梁类截面设计直接影响着车身的性能、轻量化和成本。针对车身截面“盲目设计-验证-反复修改”的正向试错设计方法，本项目提出一种截面设计反问题，实现截面“力学性能优化-几何反演生成-验证-少量修改”的逆向定量设计。针对概念设计阶段车身设计输入参数较少的问题，本项目开展了车身线性和非线性力学简化模型的研究，构建了其简化模型并验证了其精度满足工程需要；开展了截面反演机理研究，基于生成对抗网络构建了反演仿生算法构架；针对特定的车身截面复杂约束条件和轻量化目标，开展车身截面力学特性向几何信息的反演方法研究，开发了一种仿生智能反演算法—大树算法（THE ARTIFICIAL TREE ALGORITHM）；基于碰撞性能，开展了新能源汽车前纵梁关键截面研究，实现了力学性能提升和截面轻量化。通过本项目在电动汽车轻量化上的应用，有助于减少电动汽车电耗并降低电池使用量，进而降低整车成本,加快电动汽车普及发展。