一种基于神经网络的非线性飞行控制系统设计方案研究

在全状态反馈的前提下 ,设计了一种基于在线神经网络和反馈线性化的非线性直接自适应控制器。本文首先利用多重尺度摄动与动态逆技术结合 ,设计了无人驾驶飞机的解析动态逆控制器 ;然后引入一个单隐层在线神经网络来修正各种因素引起的状态误差 ,并证明了控制器的稳定性。最后对在线网络的实现做了详细描述。仿真分析表明 ,该方案具有很强的鲁棒性和对故障状态的适应性。

一种鲁棒的非线性飞行控制律设计

介绍了一种反馈线性化方法———逆系统方法 ,用以解决非线性飞行控制。针对逆系统方法存在的误差 ,介绍了一种鲁棒控制方法———基于RBF神经网络直接自适应控制方法 ,利用李亚普诺夫稳定性定理推导了神经网络权值的自适应规律 ,保证了闭环系统的稳定性。设计了针对滚转通道的神经网络 ,并应用某型号飞机进行了非故障和故障状态的仿真 ,结果证明 ,自适应神经网络控制方法补偿作用显著 。

一类非线性飞行控制系统鲁棒故障诊断

针对一类仿射非线性飞行控制系统的执行器故障,提出了一种基于微分几何方法的鲁棒故障诊断方案。各个输入通道相应的不可观分布满足条件时,可以通过分别在状态空间和输出空间进行同胚变换,将原系统化为多个只受单一通道故障输入影响的能观商子系统。原系统的残差生成器设计问题转化为对其各通道能观商子系统的观测器设计,实现残差与其他通道故障及结构性干扰解耦,从而进行鲁棒故障诊断。对不同输入通道发生的执行器故障进行仿真(包括卡死、损伤、松浮、饱和),仿真结果验证了该方法的有效性,而且可看到该方法对多故障情况也有很好的诊断效果。

基于自适应神经网络的飞行控制律设计方案

介绍了一种自适应神经网络控制方法,利用李亚普诺夫稳定性定理推导了神经网络权值的自适应规律,保证了闭环系统的稳定性。设计了针对滚转通道的神经网络,并应用某型号飞机进行了非故障和故障状态的仿真。结果表明,自适应神经网络控制方法补偿作用显著,相当于系统具有一定的重构功能。

ADAPTIVE BACKSTEPPING FLIGHT CONTROL WITH ACTUATOR SATURATION

Because actuator satu ration can become a problem or even a disaster in flight control system, the con sideration of actuator saturation in the design phase may indeed reduce the degr ee of conservativeness of an flight control system (FCS) and thus noticeably enh ance the performance of the FCS. Deflection limits and rate limits are both cons idered in a new adaptive backstepping FCS design process. The key of the method is that a new control Lyapunov function (CLF) and a control law are chosen when the actuator saturation occurs. This idea results from that there must be a vari ation in the pseudo-control at saturation. The whole progress is a modification of an early presented method: adaptive backstepping control scheme. The stabili ty is proved and verified successfully. The conclusion and some comments about t his method are given in the end.

Robust nonlinear control of spacecraft formation flying using constraint forces

A robust nonlinear control method is presented for spacecraft precise formation flying.With the constraint forces and consid-ering nonlinearity and perturbations,the problem of the formation keeping is changed to the Lagrange systems with the holonomic constraints and the differential algebraic equations (DAE).The nonlinear control laws are developed by solving the DAE.Because the traditional numerical solving methods of DAE are very sensitive to the various errors and the resulting con-trol laws are not robust in engineering application,the robust control law designed method is further developed by designing the correct coefficients to correct the errors of the formation array constraints.A numeral study simulated the robustness of this method for the various errors in the formation flying mission,including the initial errors of spacecraft formation,the reference satellite orbit determination errors,the relative perturbation forces model errors,and so on.

Nonlinear hierarchy-structured predictive control design for a generic hypersonic vehicle

A hierarchy-structured predictive controller is designed and analyzed for rotation motion dynamics of a generic hypersonic vehicle(GHV).This vehicle model has fast variability,is highly nonlinear,and includes uncertain parameters.The controller contains two subsystems,the inner-fast-loop nonlinear generable predictive controller(NGPC)and the outer-slow-loop NGPC,both of which are designed by the closed-form optimal generable predictive control method.Thus,the heavy on-line computational burden in the classical predictive control method is avoided.The hierarchy structure of the control system decreases the relative degree of each subsystem and helps increase the dynamic response speed of the attitude controller.In order to improve the robustness of the control system,a feedback correction algorithm is proposed that corrects the calculation error between the predictive model and the real dynamic model.Simulation studies are conducted for the trimmed cruise conditions of an altitude of 33.5 km and Mach 15 to investigate the responses of the vehicle to the step commands of angle of attack,sideslip angle,and bank angle.The simulation studies demonstrate that the proposed controller is robust with respect to the parametric uncertainties and atmospheric disturbance,and meets the performance requirements of GHV with acceptable control inputs.

Active control of supersonic/hypersonic aeroelastic flutter for a two-dimensional airfoil with flap

The flutter, post-flutter and active control of a two-dimensional airfoil with control surface operating in supersonic/hypersonic flight speed regions are investigated in this paper. A three-degree-of-freedom dynamic model is established, in which both the cubic nonlinear structural stiffness and the nonlinear aerodynamic load are accounted for. The third order Piston Theory is employed to derive the aerodynamic loads in the supersonic/hypersonic airflow. Nonlinear flutter happens with a phenomenon of limit cycle oscillations (LCOs) when the flight speed is less than or greater than linear critical speed. The LQR approach is employed to design a control law to increase both the linear and nonlinear critical speeds of aerodynamic flutter, and then a combined control law is proposed in order to reduce the amplitude of LCOs by adding a cubic nonlinear feedback control. The dynamic responses of the controlled system are given and used to compare with those of the uncontrolled system. Results of simulation show that the active flutter control method proposed here is effective.