基于BP神经网络对爆炸容器多层复合结构的多目标优化  

Multi-objective optimization of explosion container’s multi-layer composite structure based on BP neural network

作　　者：徐景林 夏成量 刘欣 王振雄 唐勤洪 张成卓 XU Jinglin;XIA Chengliang;LIU Xin;WANG Zhenxiong;TANG Qinghong;ZHANG Chenzhuo(Unit 31306 of PLA,Chengdu 610031,China;Rocket Force University of Engineering,PLA,Xi’an 710025,China;Chemical Defense Institute,AMS,PLA,Beijing 102205,China)

机构地区：[1]中国人民解放军31306部队,四川成都610031 [2]中国人民解放军火箭军工程大学,陕西西安710025 [3]军事科学院防化研究院,北京102205

出　　处：《防护工程》2023年第2期47-56,共10页Protective Engineering

摘　　要：利用泡沫铝材料衰减冲击波和吸收爆炸能量的特性,设计了一种泡沫铝夹心的多层复合结构爆炸容器,并利用LS-DYNA软件进行了数值模拟,得到其动态响应特性。针对多层复合结构爆炸容器的外壳应变和能量吸收指标,采用BP神经网络和NSGA-Ⅱ遗传算法对其进行了结构优化。优化结果表明:高密度的泡沫铝夹芯对于大药量爆炸载荷是合适的,低密度的泡沫铝更适于小药量情况;相同爆炸载荷作用下,内筒和泡沫铝的厚度越大,外壳应变越小;内筒和泡沫铝的厚度越小,泡沫铝的吸能效率越高;内筒厚度越小同时泡沫铝的厚度越大,整个泡沫铝夹芯吸收的总能量越多。Considering the characteristics of aluminum foam of attenuating shock wave and absorbing explosive energy,this paper designed a multi-layer composite structure explosive container sandwiched with aluminum foam,and conducted numerical simulation to it by using LS-DYNA software to obtain its dynamic response characteristics.Then,the structure was optimized using BP neural network and NSGA-II genetic algorithm in respect of two indexes of casing strain and energy absorption.The results of optimization show that:high-density aluminum foam core is suitable for large-charge explosive load,while low-density aluminum foam is more suitable for small-charge explosive load;under the same explosion load,the thicker the inner casing and the aluminum foam,the smaller the strain of the outer casing;the thinner the inner casing and the aluminum foam,the greater the energy absorption efficiency;and the thinner the inner shell and the thicker the aluminum foam,the more the total energy absorbed by the entire aluminum foam core.

关 键 词：BP神经网络 NSGA-Ⅱ遗传算法 复合爆炸容器 多目标优化 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TG146.21[自动化与计算机技术—控制科学与工程]