BP神经网络制导弹道导弹的诸元数据制备方法

An approach to data preparation of closed loop guidance based on BP neural network

采用BP神经网络逼近算法,推导、建立了预定关机点及落点坐标与需要速度间的映射关系,并装订上弹,使关机点附近对应的每一个位置都能够映射出相应的需要速度矢量,利用闭路制导关机及导引方法对导弹实施控制。提出了基于BP神经网络制导的数据制备方法,将大量的弹道数据改为以神经网络形式进行制备,有效地缩小了数据存储空间及弹载计算机的计算时间。仿真结果表明,基于BP神经网路的制导方法能够大大提高导弹的制导精度,相应的诸元数据制备方法能够准确地实现神经网络在弹上的映射功能,为该制导方法在弹上的应用奠定了基础。

Mapping relation between the burnout and droppoint coordinates and required velocity is established by using BP neural network approximation algorithm,which would be bound to the missile so that required velocity can be computed around the burnout coordinate. Then the missile is controlled by the close loop guidance law. The data preparation of BP neural network is studied,a lot of data is prepared in the form of BP neural network instead,which reduces the datastore space and computation time of projectile computer. Simulation results show that the guidance based on BP neural network can improve the ballistic missile accuracy and the data preparation scheme can realize the mapping capabilities of BP neural network on missile accurately,thus establishing the foundation for the guidance usage on missile.