摘要
利用RBF神经网络的自学习能力,并结合非线性动态逆原理,设计了一种基于RBF神经网络干扰观测器的导弹动态逆控制器。根据时标分离的原则,可将导弹系统分解为快慢不同的四个回路,本文主要对快、慢两个回路分别设计动态逆控制器,并且在慢回路利用RBF神经网络干扰观测器估计导弹所受的扰动,同时,用于在线估计动态逆误差,降低了控制器对干扰和模型精确度的要求,增强了控制器的适应性。仿真结果表明,该控制器具有有效性和鲁棒性。
For using the self - learning ability of RBF neural networks and the theory of nonlinear dynamic inversion, a dynamic inversion controller for missile with disturbance observer based on RBF neural networks is designed in the paper. According to the time scale separation principle, the missile system is divided into four subsystems. In this paper, the quick loop and the slow loop are considered. The dynamic inversion controllers are designed respectively for the quick loop and the slow loop. The disturbance observer based on neural networks is used to approximate the disturbance and to on-line compensate the dynamic inversion error. This design methed can reduce the restrain conditions on disturbance and the demand of model accuracy, Therefore, it can improve the adaptability of the controller. Simulation results show that the controller is available and has robust performance.
出处
《航空兵器》
2008年第3期36-41,共6页
Aero Weaponry
基金
航空科学基金资助项目(20075152014)
关键词
RBF神经网络
干扰观测器
动态逆
导弹控制器
RBF neural network
disturbance observer
dynamic inversion
missile controller