数据驱动的自洽聚类分析方法及塑性成形高效跨尺度模拟  

Data-driven Self-consistent Clustering Analysis and High-efficiency Cross-scale Simulation of Plastic Forming Process

作　　者：江晟达 何霁[1,2] 郭聪 李淑慧[1,2] 钱昌明[1,2] JIANG Shengda;HE Ji;GUO Cong;LI Shuhui;QIAN Changming(State Key Laboratory of Mechanical System and Vibration,Shanghai Jiao Tong University,Shanghai 200240;Shanghai Key Laboratory of Digital Manufacture for Thin-walled Structures,Shanghai Jiao Tong University,Shanghai 200240)

机构地区：[1]上海交通大学机械系统与振动全国重点实验室,上海200240 [2]上海交通大学上海市复杂薄板结构数字化制造重点实验室,上海200240

出　　处：《机械工程学报》2025年第4期74-85,共12页Journal of Mechanical Engineering

基　　金：国家自然科学基金(52475379,51790170);机械系统与振动全国重点实验室自主课题(MSVZD202207)资助项目。

摘　　要：材料微观变形决定了构件宏观成形性能,如何高效地联系宏观成形工艺和材料微观结构演化一直是塑性成形领域亟须解决的难题。通过将构件不同位置的宏观力学性能需求映射到材料的微观结构,才能根据目标微观结构的特点与分布准确设计和实施制造工艺,破解传统反复“试错”的制造模式,获得“形性”兼备的高性能构件。基于数据驱动的自洽聚类分析方法,提出跨尺度模拟单元和传统宏观唯象本构单元混合使用策略方法,开发了塑性成形高效跨尺度模拟技术。实现DP980双相钢U形件拉弯以及AA2219铝合金带筋筒体旋压的跨尺度成形分析,在能够获取成形零件关键位置微观信息演化的前提下,保证与传统宏观仿真相当的计算效率,相关研究为跨尺度模拟在工程大型构件成形中的实际应用提供了新思路。Material microstructure deformation determines macroscopic forming performance.How to efficiently link macroscopic forming processes with material microstructure evolution has always been a pressing problem in the field of plastic forming.By mapping the macroscopic mechanical performance requirements of different component positions to the material microstructures,manufacturing processes can be accurately designed and implemented based on the characteristics and distribution of the target microstructures to break the traditional trial-and-error mode and obtain high-performance components with the required shape and performance.Therefore,the strategy of mixed use of cross-scale elements and traditional macro-elements based on data-driven self-consistent clustering analysis is proposed,and the efficient cross-scale simulation technology for plastic forming is developed.The cross-scale forming analysis of DP980 U-shaped components and AA2219 ribbed cylinder is efficiently realized.The material microstructure deformation at key positions during the forming process is predicted,and the computational efficiency can be close to that of traditional macro models.This research provides new ideas for the practical application of cross-scale methods in the process design of large engineering structures.

关 键 词：塑性成形 高效跨尺度模拟 数据驱动 自洽聚类分析 

分 类 号：TG156[金属学及工艺—热处理]