摘要
提出一种基于RBF(Radial Based Function)神经网络和动态面控制的飞行控制律设计方法。针对飞机气动参数变化引起的非线性和不确定性,通过RBF神经网络在线逼近。动态面飞行控制律消除了反推设计方法中由于对虚拟控制反复求导而导致的复杂性问题。同时,在控制律设计中引入一个鲁棒因子项来补偿外界干扰和神经网络的逼近误差,提高了系统的鲁棒性,使整个系统获得更好的跟踪控制性能。基于Lyapunov稳定性定理证明了闭环系统的所有信号半全局一致终结有界,并且通过适当选择设计参数,跟踪误差可收敛到原点的一个小邻域内。最后,通过飞机俯仰运动飞行的数值仿真验证了该方法的有效性。
A design method for flight controllers using the RBF neural network and dynamic surface control is proposed.The RBF neural network is used to approximate the aerodynamic force and moment coefficients.The problem of expose in terms of traditional back stepping design,which is caused by repeated derivatives of certain nonlinear functions such as virtual control,is overcome by the dynamic surface control method.Then,a robust term is introduced to compensate the external disturbance and the approximation error,which can improve the robustness,and guarantee much higher tracking accuracy.All signals in the close loop are guaranteed to be semi-globally uniformly ultimately boundedness and the output tracking error is proved to converge to a small neighborhood around zero by appropriately choosing design parameters.Simulation results for aircraft pitch movement demonstrate the effectiveness of the proposed method.
出处
《飞行力学》
CSCD
北大核心
2010年第6期41-44,共4页
Flight Dynamics
基金
国家自然科学基金(60904038)
空军工程大学学术基金(XS0901014)
关键词
飞行控制
鲁棒性
动态面控制
神经网络
飞机俯仰运动
flight control
robustness
dynamic surface control
neural network
aircraft pitch movement
