临近空间动能拦截器神经反演姿态控制器设计

为满足临近空间动能拦截器姿态控制快速性、准确性和鲁棒性的要求，设计了一种自适应神经反演姿态控制器。首先，建立了姿控发动机侧喷干扰模型，并推导了包含质心漂移、参数摄动和外界干扰的三通道强耦合模型；其次，设计了自适应神经反演姿态控制器，为提高控制精度，采用径向基函数（RBF）神经网络对各个通道的不确定项进行估计和补偿，并基于最小学习参数的思想，将神经网络学习参数拟合为一个参数，提高了RBF计算效率，保证了估计的实时性。最后，采用伪速率（PSR）脉冲调制器将设计的连续控制律转化为脉冲控制律，实现了拦截器的变推力控制，并克服了脉冲脉宽调制（PWPF）调制器相位滞后问题。数字仿真表明，所设计的控制器收敛速度快，控制精度高，对强扰动具有鲁棒性。

Attitude control of near space kinetic kill vehicle based on neural network backtepping control

To meet the requirement for fast, accurate and robust attitude control of near space kinetic kill vehicles, an adaptive neural network backstepping attitude controller is designed. First, a model for lateral jet interaction is established, and a three-channel strong coupling model including centroid drift, parameter perturbation and external disturbance is developed. Second, an adaptive neural backstepping attitude controller is designed. Then, to improve control precision, the RBF neural network is adopted to estimate and compensate the uncertainties in each channel, and the neural network learning parameter is fitted as a parameter based on the theory of minimum learning parameters to improve Radial Basis Function (RBF) calculation efficiency and estimation instantaneity. Finally, a PSeudo Rate (PSR) modulator is designed to convert the continuous control law into the pulse control law, so as to realize control of KKV digital variable thrust. The simulation shows that the controller has fast convergence speed, high control precision and robustness to strong disturbance.