基于BP神经网络预测炸药JWL状态方程参数对EFP速度的影响  

Prediction of the influence of parameters of JWL equation of state on EFP velocity based on BP neural network

作　　者：郝礼楷 顾文彬 谢兴博 张亚栋 钟明寿 邹绍昕 康耕新 HAO Likai;GU Wenbin;XIE Xingbo;ZHANG Yadong;ZHONG Mingshou;ZOU Shaoxin;KANG Gengxin(College of Field Engineering,Army Engineering University of PLA,Nanjing 210000,China;Unit 31539 of the Chinese People’s Liberation Army,Beijing 100000,China;Engineering Research Center of Explosion Safety and Protection,Ministry of Education,Southeast University,Nanjing 210000,China;National Defense Engineering College,Army Engineering University of PLA,Nanjing 210000,China)

机构地区：[1]陆军工程大学野战工程学院,南京210000 [2]中国人民解放军31539部队,北京100000 [3]东南大学爆炸安全防护教育部工程研究中心,南京210000 [4]陆军工程大学国防工程学院,南京210000

出　　处：《兵器装备工程学报》2023年第6期47-55,共9页Journal of Ordnance Equipment Engineering

基　　金：科技创新项目(KYGYZB0019003)。

摘　　要：为达成快速精准预测爆炸成型弹丸(explosively formed projectile,EFP)速度以及检验校准JWL状态方程参数等目的,使用LS-DYNA有限元软件建立Φ65 mm EFP的数值模型,研究JWL状态方程各参数对EFP速度的影响规律。在此基础上,拟合建立基于JWL状态方程参数的EFP速度计算公式,构建EFP速度的BP神经网络预测模型,生成学习样本对网络模型进行训练。结果表明,JWL状态方程参数对EFP速度存在显著影响,参数R_(1)影响最大,各参数影响的大小顺序为R_(1)、A、B、R_(2)、E_(0)、ω;计算公式和网络模型均能够精准预测EFP速度,网络模型的预测精度优于计算公式,其中计算公式的结果误差绝大部分能够控制在10%以内,网络模型的预测误差均不超过5%。To quickly and accurately predict the velocity of an explosively formed projectile(EFP)and calibrate the parameters of the JWL equation of state,this paper uses LS-DYNA finite element software to establish a numerical model ofΦ65 mm EFP,and analyzes the influence law of the parameters of the JWL equation of state on EFP velocity.On this basis,the calculation formula of the EFP velocity is fitted based on the parameters of the JWL equation of state,a BP neural network prediction model of the EFP velocity is constructed,and the learning samples are generated to train the network model.The results show that the parameters of the JWL equation of state have a great influence on the EFP velocity.R_(1)has the greatest influence and the order of the influence of each parameter is R_(1),A,B,R_(2),E_(0)andω.Both the calculation formula and the network model can accurately predict the EFP velocity,and the prediction accuracy of the latter is better than that of the former.Most of the errors of the calculation formula can be controlled within 10%,and the prediction errors of the network model are not more than 5%.

关 键 词：JWL状态方程 神经网络 爆炸成型弹丸(EFP) 速度 

分 类 号：TJ55[兵器科学与技术—军事化学与烟火技术] O389[理学—流体力学]