一种神经网络弹道诸元快速解算方法  

A Fast Calculation Method for Solving Firing Data by Neural Network

作　　者：连欢 邓泽晓 李志国 黄祺威 刘鲁华 LIAN Huan;DENG Zexiao;LI Zhiguo;HUANG Qiwei;LIU Luhua(Shenzhen Key Laboratory of Intelligent Microsatellite Constellation,Shenzhen 518107,China;Beijing Institute of Astronautical Systems Engineering,Beijing 100076,China)

机构地区：[1]深圳市智能微小卫星星座技术与应用重点实验室,深圳518107 [2]北京宇航系统研究所,北京100076

出　　处：《航天控制》2024年第5期38-44,共7页Aerospace Control

基　　金：国家自然科学基金资助(61973326);深圳市科技计划资助(ZDSYS20210623091808026)。

摘　　要：针对导弹快速发射的任务需求,提出一种神经网络诸元快速解算方法。首先,根据任务需求建立了终端速度、高度、速度倾角和弹道诸元的映射关系,推导了基于LM优化算法的神经网络参数更新方法;然后,基于贝叶斯正则化理论设计了BP神经网络结构,得到满足精度要求的优化网络结构;最后,利用牛顿迭代法生成诸元数据库,将其作为训练集对神经网络进行训练,获得了具有优化参数的网络模型,并展开了仿真验证。理论与仿真结果表明,该方法可以实现射前诸元的快速精确计算,有效缩短射前准备时间。Regarding meeting the task requirement of rapid missile launch,a fast firing data solving method based on neural network is proposed.Firstly,the mapping relationship among terminal velocity,altitude,velocity inclination and firing data is established,and the LM optimization neural network model is derived.Then,based on Bayesian regularization theory,the BP neural network structure is designed,and the optimized network structure which can meet the accuracy requirements is obtained.Finally,Newton iteration method is used to generate the database as training set which is used to train the neural network,so that the network model with optimized parameters is obtained,and the simulation is implemented.The theoretical and simulation results show that the rapid calculation of the firing data before shooting can be achieved by applying this method.

关 键 词：弹道诸元 LM优化算法 BP神经网络 贝叶斯正则化 

分 类 号：TJ760[兵器科学与技术—武器系统与运用工程]