基于极限学习机的杀爆战斗部末弹道参数优化研究

Study on Optimization of Final Trajectory Parameters of Blasting Warheads Based on Extreme Learning Machine

为研究末弹道参数与杀伤面积之间的复杂关系,以达到优化末弹道参数的目的,对破片飞散过程进行分析,为描述破片飞行轨迹建立了破片飞散模型,得到破片的落点,然后通过统计地面有效破片数,求解战斗部在不同末弹道情况下的杀伤面积,得到1225种末弹道组合下的杀伤面积,最后基于这些数据,利用极限学习机(ELM)得到1903993种末弹道组合的计算结果。计算结果表明:当极限学习机的激励函数采用sigmoid函数,隐藏节点数在200个以上时,决定系数能够达到0.9以上,求解时间远远小于1 s,求解得到的最优末弹道参数对应的杀伤面积从优化前的438.1 m^(2)上升至优化后的541.2 m^(2)。

In order to study the complex interaction of the final trajectory parameters and the lethal area to optimize the final trajectory parameters,the fragment dispersion process was analyzed,and the fragment dispersion model was established to calculate the effective fragments on the ground.The lethal area of the warhead under different final trajectory parameters was solved by counting the number of effective fragments on the ground.The lethal area under 1225 final trajectory combinations was obtained.On this basis,the extreme learning machine(ELM)was used to optimize the calculation of 1903993 final trajectory combinations.The results show that the determination coefficient can reach more than 0.9,and the solution time is much less than 1 s,and the lethal area increases from 438.1 m^(2) to 541.2 m^(2) when the number of hidden nodes is more than 200 and the sigmoid function is used to the extreme learning machine.