首页|期刊导航|应用数学和力学|小样本数据驱动与约束融合的多目标拓扑优化方法

小样本数据驱动与约束融合的多目标拓扑优化方法OA

A Small Sample Data Driven and Constraint Fusion Multi-Objective Topology Optimization Method

中文摘要英文摘要

传统体积-柔顺性的多目标拓扑优化方法依赖密集有限元计算,存在计算效率低、Pareto 解集覆盖稀疏等问题.本文提出了一种基于小样本数据驱动与约束融合的高效多目标拓扑优化方法.该方法的核心是建立含物理约束嵌入的小样本条件生成对抗网络,学习少量有限元计算次数下的拓扑结构数据集,生成高精度、完整的多目标拓扑优化 Pareto 最优解集.该方法通过小样本数据驱动的方式构建体积分数和多目标拓扑结构之间的关系,在保证计算精度的同时降低对大规模有限元计算的依赖.并且多种物理约束融合保证物理场-拓扑高精度映射,为多目标优化设计提供了高效且可靠的解决方案.

The traditional volume-compliance multi-objective topology optimization methods rely on intensive finite element calculations,often suffering from low computational efficiency and sparse coverage of the Pareto solution set.To address these issues,an efficient multi-objective topology optimization method based on small sample data driving and constraint fusion was proposed.The core of this method is the establishment of a small sample conditional generative adversarial network embedded with physical constraints,which learns from a limited number of finite element calculations to generate a high-precision and comprehensive Pareto optimal so-lution set for multi-objective topology optimization.By means of the small sample data driven techniques,the relationships between volume fractions and multi-objective topological structures were established.The results show that,the proposed method ensures computational accuracy while minimizing reliance on large-scale finite element calculations.Furthermore,the fusion of multiple physical constraints guarantees high precision map-ping between physical fields and topologies,providing an efficient and reliable solution for multi-objective opti-mization design.

应佳桓;周焕林;孟增

合肥工业大学 土木与水利工程学院,合肥 230009合肥工业大学 土木与水利工程学院,合肥 230009合肥工业大学 土木与水利工程学院,合肥 230009

通用工业技术

多目标拓扑优化小样本数据驱动物理约束生成对抗网络

multi-objective topology optimizationsmall sample datadata drivenphysical constraintgener-ative adversarial network

《应用数学和力学》 2026 (9)

1107-1116,10

国家自然科学基金(1257222412372195)

10.21656/1000-0887.460112

评论