论文部分内容阅读
论述了神经网络理论在飞行器再入制导方面的应用。在分析各方法优缺点的基础上,提出了一种基于广义回归神经网络(GRNN)模型的再入制导方法。神经网络通过一组选定的轨迹样本进行有导师训练,训练好之后作为控制指令生成器,输入为再入过程中飞行器的状态参数,输出为倾侧角控制量,迎角控制量则由迎角剖面给定。仿真结果验证了该方法在再入飞行器存在再入初态误差和损伤情况下的可行性及鲁棒性,具有良好的工程应用前景。
The application of neural network theory in aircraft reentry guidance is discussed. Based on the analysis of the advantages and disadvantages of each method, a re-entry guidance method based on GRNN model is proposed. The neural network conducts mentor training through a set of selected trajectory samples. After training, it is used as a control command generator, input as the state parameter of the aircraft during reentry process, the output is the roll angle control amount, and the angle of attack control amount is from angle of attack Profile given. The simulation results verify the feasibility and robustness of this method in the case of reentry initial state error and damage of reentry vehicle and have good engineering application prospect.