摘要
在建立某型火箭炮动力学模型的基础上,根据正交试验的原则,通过动力学仿真和数据处理为BP网络建立训练样本,用训练后的网络模拟发射间隔和起始扰动之间的非线性关系,将改进后的自适应遗传算法(IAGA)和BP网络结合对发射间隔进行研究和优化,得出了变发射间隔的满意解。结果表明,将BP和IAGA结合,既克服了BP优化功能的不足,又弥补了遗传算法优化时需要显式目标函数的缺陷,解决了单纯用动力学仿真不能解决的问题。优化的结果可以直接应用到该型火箭炮的发射中去。
On the base of the establishment of a certain rocket launcher model, some samples for training the BP neural network were got by using an orthogonal experimental method through the dynamical simulation. The trained neural network could simulate the nonlinear relation between firing interval and initial disturbance. The firing interval was studied and optimized to obtain a reasonable result using the improved adaptive genetic algorithm(IAGA) in conjunction with BP neural network. The results indicate that the cooperation of BP and IAGA can resolve a certain question which is not successfully resolved simply using the dynamical simulation. The optimized result can be used in the firing of a certain rocket launcher.
出处
《兵工学报》
EI
CAS
CSCD
北大核心
2007年第11期1287-1292,共6页
Acta Armamentarii
关键词
机械学
BP网络
遗传算法
自适应
变发射间隔
优化
火箭炮
mechanics
BP neural network
genetic algorithm
adaptive
variable firing interval
optimization
rocket launcher