基于代理模型技术的玻璃纤维复合装甲抗弹性能多目标优化设计  

Multi-objective optimization of the ballistic resistance of a glass fiber reinforced composite armor based on a surrogate model

作　　者：王威 胡鹏程 张攀[1] 刘均[1] 程远胜[1] WANG Wei;HU Pengcheng;ZHANG Pan;LIU Jun;CHENG Yuansheng(School of Naval Architecture and Ocean Engineering,Huazhong University of Science and Technology,Wuhan 430074,China)

机构地区：[1]华中科技大学船舶与海洋工程学院,武汉430074

出　　处：《振动与冲击》2022年第18期16-24,共9页Journal of Vibration and Shock

摘　　要：针对一种由玻璃纤维复合材料和高强钢组成的复合装甲结构,开展其在圆柱形弹体侵彻下的弹道行为研究。基于验证后的数值模型,采用Kriging代理模型技术,建立了复合装甲结构抗弹性能快速预报代理模型,检验了代理模型的预报精度,并开展了变量相关性分析。建立了玻璃纤维复合装甲抗弹性能多目标优化设计数学模型,采用NSGA-Ⅱ多目标遗传算法,获得了复合装甲结构多目标优化设计帕累托前沿。结果表明:弹道极限速度与芯层厚度的线性相关性最强,与后置面板厚度的线性相关性最弱;靶板面密度与芯层厚度的线性相关性最强,与前后面板厚度的线性相关性相等。针对典型复合装甲结构初始设计方案,通过开展多目标优化设计,优化方案能够在相同抗弹性能的前提下使面密度降低15.64%,或者在相同面密度水平的前提下使弹道极限速度提高14.75%。The ballistic behavior of a novel composite armor consisting of high strength steel panels and a glass fiber reinforced polymer core was investigated numerically.Based on the verified numerical model,a surrogate model with good accuracy was constructed to rapidly predict the ballistic performance of the composite armor using the Kriging technique.A correlation analysis among the thickness of each layer,the areal density and the ballistic limit was conducted.Finally,a mathematical model was established for the multi-objective optimization design of the composite armor.The corresponding Pareto front of the composite armor was achieved using the NSGA-Ⅱ multi-objective genetic algorithm.The results show that the ballistic limit has the strongest correlation with the core thickness and the weakest correlation with the rear panel thickness.The areal density has the strongest correlation with the core thickness and the identical correlation with the thickness of the two face sheets.Through the multi-objective optimization design,the initial design of the composite armor gains 15.64% reduction in the areal density or 14.75% increase in the ballistic limit with the other index keeping the same level.

关 键 词：复合装甲 玻璃纤维增强材料 数值仿真 代理模型 多目标优化 

分 类 号：TH212[机械工程—机械制造及自动化] TH213.3