基于网格划分与BP网络的中制导弹道在线生成方法

Trajectory Generation for Midcourse Guidance Based on Grid Partition and BP Neural Network

针对临近空间高超声速目标高速、高机动的特点,为改善拦截弹中制导性能,将基于网格划分的中制导离线弹道优化与神经网络相结合,提出了一种中制导弹道在线生成方法。以拦截区域网格化形成的网格点为约束,采用离线弹道优化方法,形成覆盖拦截区域的弹道数据库;利用BP神经网络对非线性函数的逼近特性,通过训练实现对弹道数据库的逼近,构建BP神经网络模型,实现中制导弹道在线生成。仿真分析结果表明:中制导过程中,在终端约束调整和不调整的2种情况下,该文设计的中制导弹道生成方法均具有较好的性能。

Aiming at the characteristics of high speed and high maneuvering of near space hypersonic- target ,the midcourse guidance off-line trajectory optimization based on grid division was combined with neural network,and an on-line trajectory generation method for midcourse guidance was proposed to improve the midcourse-guidance performance of interceptor missile. Taking the grid points obtained by meshing the intercept area as constrain,the trajectory data base covering intercept area was constructed by means of off-line trajectory optimization. Based on the property of BP neural network approximating any nonlinear functions,the approximation of trajectory data base was achieved ,and the BP neural network model was constructed. The on-line generation of midcourse-guidance trajectory was realized. Simulation results show that in the process of midcourse guidance,the proposed generation method of midcourse- guidance trajectory has good performance with and without adjusting terminal constrain.