航空学报 > 2013, Vol. 34 Issue (12): 2645-2657   doi: 10.7527/S1000-6893.2013.0221

基于气动特性辨识的飞行器抗饱和自适应控制

王超, 张胜修, 郑建飞, 张超   

  1. 第二炮兵工程大学 自动控制工程系, 陕西 西安 710025
  • 收稿日期:2013-01-11 修回日期:2013-04-13 出版日期:2013-12-25 发布日期:2013-05-07
  • 通讯作者: 张胜修,Tel.:029-84741963E-mail:zsx1963@aliyun.com E-mail:zsx1963@aliyun.com
  • 作者简介:王超男,博士研究生。主要研究方向:模型预测控制、精确制导与控制。Tel:029-84741963E-mail:dieche1218@sina.com;张胜修男,博士,教授,博士生导师。主要研究方向:组合导航与飞行器制导控制。Tel:029-84741963E-mail:zsx1963@aliyun.com;郑建飞男,博士研究生。主要研究方向:飞行器控制、仿真与决策。Tel:029-84741963E-mail:zjf1981@163.com;张超男,博士研究生。主要研究方向:飞行器控制、仿真与决策。Tel:029-84741963E-mail:zc1983@gmail.com
  • 基金资助:

    国家自然科学基金(61203354)

Anti-windup Adaptive Control of Aircraft Based on Online Identification of Aerodynamic Characteristics

WANG Chao, ZHANG Shengxiu, ZHENG Jianfei, ZHANG Chao   

  1. Department of Automatic Control Engineering, the Second Artillery Engineering University, Xi'an 710025, China
  • Received:2013-01-11 Revised:2013-04-13 Online:2013-12-25 Published:2013-05-07

摘要:

针对无人机含有未知气动参数时的大机动飞行控制问题,设计了基于在线气动参数辨识的自适应非线性模型预测控制(ANMPC)的飞行控制器。首先,根据无人机动力学模型设计出基于反馈线性化模型的非线性模型预测控制(NMPC)的飞行控制器,通过在线求解有约束的最优化问题,确保大机动飞行时系统状态及控制输入满足约束条件,使控制系统具备抗饱和性能。然后,将无人机动力学模型中的未知气动参数转换为待辨识参数矩阵,采用基于迭代扩展卡尔曼滤波(IEKF)与渐消记忆最小二乘(RLS)的气动特性辨识两步方法,实时辨识无人机气动参数,更新非线性模型预测控制律所用模型,消除模型误差,增强非线性模型预测控制器的鲁棒性。最后,对指令姿态角跟踪进行了仿真验证,仿真结果表明:在考虑外界未知气动扰动情况下,控制器满足设计要求,并具有较强的鲁棒性。

关键词: 飞行控制, 气动扰动, 气动特性辨识, 自适应控制, 非线性模型预测控制, 反馈线性化

Abstract:

A high maneuver flight controller using online identification of aerodynamic parameters based adaptive nonlinear model predictive control (ANMPC) is proposed for an unmanned aerial vehicle with unknown aerodynamic parameters. Firstly, an nonlinear model predictive control (NMPC) controller is designed in accordance with the feedback-linearized unmanned aerial vehicle dynamical model. The systematic handling of input and state constraint violations in a high maneuvering flight is achieved by solving an online constrained optimization problem. In this way, accomplishment of anti-windup is guaranteed. Secondly, the unknown parameter matrixes, which are formed from the unknown aerodynamic parameters in the unmanned aerial vehicle dynamic model, are identified in real time by means of a two-step method which combines the iterated extended Kalman filter (IEKF) and the recursive least square (RLS) estimation with forgetting. In the meantime, the unmanned aerial vehicle dynamical model is updated with identified aerodynamic parameters to compensate for model mismatch. In this way, robust performance of the NMPC controller is improved with updated model information. Finally, simulation results for attitude command tracking in the presence of perturbed aerodynamic parameters demonstrate that all perfor-mance requirements are satisfied and the robustness is successfully achieved by the designed controller.

Key words: flight control, aerodynamic disturbance, identification of aerodynamic characteristic, adaptive control, nonlinear model predictive control, feedback linearization

中图分类号: