基于神经网络的精确末制导律研究

针对动能杀伤武器重量极小化问题，基于神经网络技术、动态逆补偿法和非线性优化法，研究了两种精确末制导律，一种采用FFBP神经网络学习一要制导方程动态逆实现精神制导，另一种采用FFBP神经网络学习非线性优化解实现精确制导，仿真分析表明，前者对目标加速度估计误差过于敏感，无法实现；而后者在目标加速度有误差时仍能实现精确制导，估于偏置比例导引，并可以使动能杀伤武器重量极小化。

Exact terminal guidance using neural networks

To miniaturize the weight of kinetic energy kill vehicles, it is demonstrated that using a neural network for terminal guidance law synthesis can reduce the interceptor's weight when the divert incremental velocity change along the interceptor trajectory is minimized while the neural network dynamic inversion compensation is unpractical because it's too sensitive to target acceleration estimation errors. Comparing with the augmented proportional navigation missing target, the neural network guidance law based on numerical optimization solution yields increased weight savings.