基于优化弹道的BPNN预测制导方法

BPNN Predictor-corrector Guidance Based on Optimized Trajectory

针对飞行器再入滑翔过程,提出一种跟踪优化弹道的BPNN(BP neural network)预测制导方法。首先着眼于多约束下弹道生成的快速性,利用hp-自适应伪谱法进行弹道优化;然后利用弹道样本数据训练BPNN,建立飞行状态参数与终端状态参数之间的非线性映射关系,实现对终端状态的预测;最后为制导律设计了双层线性反馈校正算法,从而完成预测制导关键环节。仿真算例该表明制导方法能够良好地满足再入飞行约束和终端约束,同时可以较好地实现对优化弹道的跟踪,并具有一定的鲁棒性和航程适应性。

A BPNN predictor-corrector guidance method that tracks the optimized trajectory of hypersonic reentry glide process is presented. First, based on the trajectory generation rapidity under multiple constraints, the hp-adaptive pseudospectral method is used to optimize the trajectory. Then a BPNN （BP neural network） is trained with parameter data of optimized trajectory to simulate the nonlinear mapping relationship between the current flight states and terminal states. Hence, the guidance is achieved by nullifying the terminal errors. Simulation examples show that the guidance method based on trajectory optimization and neural network can satisfy both reentry flight and terminal constraints and have good robustness and validity.