基于神经网络的中制导律

基于神经网络的中制导律适合末段采用寻的制导的复合制导导弹。采用针对固定目标、同时能满足复合制导中制导要求的最优中制导律来提供神经网络的训练样本，通过遗传算法对神经网络的各个权值及各神经元的阈值进行综合寻优，以各时刻导弹及目标的位置坐标作为神经网络输入，经过计算输出相应的中制导指令。仿真结果表明基于神经网络的中制导律不仅能用于攻击固定目标及机动目标，同时能满足复合制导中末交班的各项要求。

MIDCOURSE GUIDANCE LAW BASED ON NEURAL NETWORK

A kind of midcourse guidance law was studied based on neural network. The midcourse guidance law is suitable to be used in combined guidance in which proportional navigation law is used in the terminal guidance. Training samples were provided by designing optimal guidance law which can meet the handover conditions between midcourse guidance and terminal guidance off line. Each weight of NN and threshold of neural cell are optimized through genertic algorithm. The position coordinate of missile and target is used as input of NN, the respective midcourse guidance instruction used as output after computation of NN. The simulation results show that midcourse guidance law based on neural network can be used to attack both fixed targets and maneuverable targets and also meet the handover conditions between midcourse guidance and terminal guidance.