基于神经网络的直升机动态逆飞行控制  

Helicopter dynamic inversion flight control based on neural network

作　　者：刘晶[1] 吴爱国[1] 郭润夏[2] 方星[1] 

机构地区：[1]天津大学电气与自动化工程学院,天津300072 [2]中国民航大学航空自动化学院,天津300300

出　　处：《飞行力学》2014年第3期218-221,共4页Flight Dynamics

基　　金：国家自然科学基金资助(61170328)

摘　　要：传统直升机动态逆控制器包含标称系统动态线性化部分,以及用于实现期望闭环动态的线性控制部分。针对其依赖于直升机精确非线性模型而鲁棒性较差的问题,提出了一种改进的黑鹰UH-60直升机动态逆控制方法,其中逆模型为基于悬停飞行条件下的线性模型。对模型简化、非线性和飞行条件差异所产生的逆误差,引入在线神经网络进行补偿并对系统稳定性进行了分析。非线性仿真结果表明,在不同飞行条件产生的扰动作用下,系统输出响应能够跟踪指定输入,进而说明控制器具有良好的鲁棒性。Classical helicopter dynamic inversion controller consists of a part that linearizes the nominalsystem dynamics, and a linear part that imposes the desired closed-loop dynamics. Consider the poor ro-bustness due to dependence on exact availability of the helicopter nonlinear model, this paper presents amodified dynamic inversion control method for Black Hawk UH-60 and the inversion model is based onthe linearized aerodynamics of hover. The neural network＇ s ability is to reconstruct and cancel the errorscaused by simplifications, non-linearity, and variations in operating conditions. Then stability of theflight controller is proved. The results of nonlinear simulations show that the responses can track the pilot＇s command as the flight condition changes. And the robustness of flight controller is realized.

关 键 词：直升机 动态逆 在线神经网络 逆误差 

分 类 号：V249.1[航空宇航科学与技术—飞行器设计] V275.1