近空间飞行器鲁棒自适应backstepping设计  

Robust Adaptive Backstepping Design for Near Space Vehicle

作　　者：张强[1,2] 吴庆宪[1] 姜长生[1] 王玉惠[1] 

机构地区：[1]南京航空航天大学自动化学院,江苏南京210016 [2]济南大学自动化与电气工程学院,山东济南250022

出　　处：《控制工程》2013年第2期204-208,共5页Control Engineering of China

基　　金：国家自然科学基金(90716028;60974106);国家自然青年科学基金(11102080);济南大学科研基金(XKY0916);航空科学基金(20095152028);南京航空航天大学基本科研业务费专项科研项目资助(NS2010077)

摘　　要：针对变后掠翼近空间飞行器(near space vehicle,NSV)在大包络、多任务模式飞行运动过程中具有非线性、快时变、强耦合和不确定的特性,提出了基于径向基神经网络(radialbasis function neural network,RBFNN)的鲁棒自适应跟踪控制策略。首先,利用RBFNN在线逼近NSV飞行过程中外部干扰。其次,应用backstepping设计光滑的反馈控制器。其中,采用微分器避免backstepping设计中出现微分膨胀问题,利用鲁棒项减少RBFNN估计误差对系统的影响。然后,通过公共Lyapunov函数证明所提出的控制器可以保证在任意飞行模态中NSV的输出跟踪误差均可以收敛到任意小的有界集内。最后,仿真结果表明该飞控系统具有良好的控制性能。The variable-sweep wings near space vehicle （NSV） has nonlinearity, time variability, coupling and uncertainty, espe cially in large flight envelop and multi-task modes. So based on radial basis function neural network （RBFNN）, robust adaptive track ing control is presented. Firstly, RBFNN are employed to compensate for unknown disturbance in the process of flight. Secondly, a- daptive backstepping method is then used to design the smooth feedback controller, in which the differentiator is used to avoid differen tial expansion problems in the backstepping design and robust term is to reduce the impact of estimation error from the RBFNN. Then, by using the Common Lyapunov function approach, we prove that the tracking error in the different flight modes can converges to an ar bitrary small neighbourhood of the desired trajectory. Finally, the simulation results are presented to show the effectiveness of the pro posed scheme.

关 键 词：切换非线性系统 回馈递推 自适应神经网络 近空间飞行器 

分 类 号：TP27[自动化与计算机技术—检测技术与自动化装置]