用结构自适应前馈网络解算多目标导弹攻击区

多目标导弹攻击区的实时解算是机载多目标攻击武器系统得以技术实现的中心环节。阐述了“多目标导弹攻击区”的概念 ,探索了其解算的技术途径 ,提出了“实时、高精度和低存储”的解算要求。根据上述技术要求 ,设计了基于网络灵敏度统计分析的多层前馈网络结构自适应算法 ,以用于训练合适规模的多层前馈网络逼近器。在此基础上 ,进行了系统仿真研究 ,研究的结果表明了该方案的应用前景。

Calculating Missile Launching Envelope in Multi-target Attacking by Using Structural Self-adaptive Feed-forward Neural Networks

Real time processing of multi target missile launching envelope is the central link in airborne multi target attacking system.The concept of multi target missile launching envelope in multi target attacking(MTA MLE) is defined at first,and how to calculate MTA MLE is developed in this paper.According to these technical demands,a self adaptive algorithm based on network sensitivity of multi layer feed forward meural networks(MFNN) is given that is used to train a network approcher with certain scale.Based on these,some simulations are carried out,the results are promising.