高超声速飞行器的神经网络PID控制  

Neural Network PID Control of Hypersonic Vehicle

作　　者：管萍[1] 和志伟[1] 戈新生[1] 

机构地区：[1]北京信息科技大学自动化学院,北京100192

出　　处：《航天控制》2018年第1期8-13,共6页Aerospace Control

基　　金：国家自然科学基金(11472058)

摘　　要：高超声速飞行器是一种强耦合性、存在不确定性和高度非线性的复杂控制系统,在其应用神经网络PID控制方法进行纵向通道的姿态控制时,将控制系统分为快慢内外环,基于神经网络控制理论,针对高超声速飞行器的动力学模型和运动学模型设计单神经元PID控制律和PID神经网络控制律,推导了权值调整的学习算法。仿真结果表明,高超声速飞行器的神经网络PID控制系统能够有效抑制气动参数变化的影响,较好地跟踪期望攻角,有较强的鲁棒性。Hypersonic vehicle is a system with strong coupling, large uncertainty and highly nonlinear. The neural network PID control method is applied to hypersonic vehicle longitudinal channel attitude con- trol. By dividing the system into inner and outer cycle, based on the neural network control theory, the sin- gle neuron PID control algorithm and the PID neural network control algorithm are designed for the hypersonic vehicle dynamics model and kinematics model, and the parameter adjustment learning algorithm is deduced. The simulation results show that the neural network PID control system of hypersonic vehicle can restrain the influence of the change of aerodynamic parameters effectively, track the expected pitch angle quickly and accurately and has strong robustness.

关 键 词：高超声速飞行器 姿态控制 单神经元PID控制 PID神经网络控制 

分 类 号：V249.122.2[航空宇航科学与技术—飞行器设计]