Page 21 - 《爆炸与冲击》2026年第5期
P. 21
第 46 卷 第 5 期 爆 炸 与 冲 击 Vol. 46, No. 5
2026 年 5 月 EXPLOSION AND SHOCK WAVES May, 2026
DOI:10.11883/bzycj-2025-0103
基于 GNN/KAN 的高应变速率金属材料
本构关系的表征方法 *
1
1
袁基宸 ,黄夏旭 ,解国良 2
(1. 北京科技大学机械工程学院,北京 100083;
2. 北京科技大学新金属材料全国重点实验室,北京 100083)
摘 要 : 为 准 确 表 征 金 属 材 料 在 高 应 变 速 率 下 的 应 力 -应 变 本 构 关 系 , 提 出 了 基 于 图 神 经 网 络 ( graph neural
networks,GNN)和 KAN(Kolmogorov-Arnold networks)的本构关系的高精度预测模型。为解决传统 Johnson-Cook(JC)模
型不考虑温度、应变速率与应变之间的耦合效应问题,在 GNN 模型中构建图结构数据以描述多维参数的非线性关联,
在 KAN 模型中基于 Kolmogorov-Arnold 定理实现高维输入空间的非线性映射。基于 ODS(oxide dispersion strengthened)
铜合金的高应变率压缩实验,评估了 GNN、KAN 和 JC 的本构关系描述和预测精度。结果表明:GNN 与 KAN 模型在测
试集中的平均相对误差分别为 8.0% 与 9.0%,决定系数均高于 0.95,显著优于 JC 模型(平均相对误差为 38.0%,决定系数
为 0.75);将所构建的本构关系模型应用在有限元仿真中,GNN 和 KAN 模型预测的等效塑性应变与应力分布更符合理
论特征,而 JC 模型无法准确描述材料的软化阶段,仿真结果偏差较大。所构建的模型能有效捕捉高应变速率下材料
的多场耦合特性,为极端载荷条件下的应力-应变本构关系提供了新的预测方法。
关键词: 深度学习;高应变速率;本构关系;图神经网络;KAN;动态力学性能预测
中图分类号: O347.3 国标学科代码: 13015 文献标志码: A
Characterization method of material constitutive relationship
at high strain rates based on GNN/KAN
1
1
YUAN Jichen , HUANG Xiaxu , XIE Guoliang 2
(1. School of Mechanical Engineering, University of Science and Technology Beijing, Beijing 100083, China;
2. National Key Laboratory of New Metal Materials, University of Science and Technology Beijing, Beijing 100083, China)
Abstract: To accurately characterize the stress-strain constitutive relationship of metal materials under high strain-rate
conditions, a novel, high-precision constitutive-relationship-prediction model based on Graph Neural Networks (GNNs) and
Kolmogorov-Arnold Networks (KANs) was developed. Traditional Johnson-Cook (JC) models often fail to account for the
coupling effects among temperature, strain rate, and strain, all of which are crucial for describing the dynamic behavior of
materials under extreme conditions. This limitation was addressed by constructing graph-structured data in the GNN model to
capture the nonlinear correlations of multidimensional parameters and by leveraging the Kolmogorov-Arnold theorem in the
KAN model to achieve precise mapping of high-dimensional input spaces. The research methodology involved several key
steps. Experimental data from ODS copper subjected to high-strain-rate compression were collected using a split Hopkinson
pressure bar (SHPB) system and subsequently preprocessed. The dataset included temperature, strain rate, strain, and stress. In
the GNN model, when temperature and strain rate were held constant, nodes were connected sequentially based on strain
values to form edges. When temperature was held constant, a reasonable threshold was established between nodes with
* 收稿日期: 2025-04-02;修回日期: 2025-07-01
基金项目: 国家重点研发计划(2021YFB3700700)
第一作者: 袁基宸(2002- ),男,硕士,M202320731@xs.ustb.edu.cn
通信作者: 黄夏旭(1985- ),男,博士,副教授,huangxx@ustb.edu.cn
051421-1

