基于机器学习的仿生点阵结构抗压强度预测系统设计与实现
摘要
与工程适用性,构建了一套融合多模型集成学习与工程可视化的智能预测系统。本系统基于81组全因子有限元仿真
数据,训练岭回归、随机森林、XGBoost及堆叠集成学习模型,仅需输入杆径、杆长、夹角与倾角四个参数,即可
实现毫秒级强度预测。同时提出应力+应变本构关系参数化生成算法,结合空间坐标变换实现三维结构可视化。集
成用户管理、批量预测与历史数据检索功能。实验表明:Stacking集成模型预测性能最优,决定系数0.985,均方根
误差21.3 MPa,平均绝对误差16.8 MPa;单点预测响应0.28秒,20样本批量预测耗时1.1秒,可在全平台运行,为航
空航天等领域轻量化结构快速设计提供高效工程支持。
关键词
全文:
PDF参考
[1]FERNANDES M C, SHARMA A, ZHANG Y,
et al. Mechanically robust lattices inspired by deep-sea glass
sponges[J]. Nature Materials, 2021, 20(2): 237-241.
[2]TUNINETTI V, ZHANG Z, WANG Y, et
al. Biomimetic lattice structures: Design principles and
manufacturing advances[J].Biomimetics, 2025, 10(7): 458.
[3]谢炜豪,杨永泰. 光固化制造仿生蝴蝶点阵结构
压缩及吸能性能研究[J]. 应用力学学报,2024,41(10):
124-132.
XIE W H, YANG Y T. Compressive and energy
absorption properties of biomimetic butterfly lattice structures
fabricated by stereolithography[J]. Chinese Journal of Applied
Mechanics, 2024, 41(10): 124-132. (in Chinese)
[4]GIBSON L J, ASHBY M F. Cellular solids: Structure
and properties[M]. 2nd ed. Cambridge: Cambridge University
Press, 1999.
[5]XIANG Y, LI H, CHEN W, et al. A GANbased stepwise full-field mechanical prediction model for
metamaterials[J]. International Journal of Mechanical Sciences,
2024, 284: 109771.
[6]NGUYEN T, PHAM H, TRAN M, et al. Bayesian
optimization of stacking ensemble learning model for concrete
strength prediction[J]. Expert Systems with Applications,
2025, 259: 125175.
Refbacks
- 当前没有refback。
