高超声速飞行器输入受限自适应反演控制研究

Study on Adaptive Back-Stepping Control of Hypersonic Vehicles with Input Constraints

针对输入受限和参数不确定时的高超声速飞行器控制问题,提出了一种基于径向基函数(RBF)神经网络补偿的自适应反演控制方法。建立了飞行器纵向运动模型,分析了由控制系统执行机构、弹性振动和避免发动机燃烧室热雍塞导致的燃料-空气比和升降舵偏角受限,通过设计辅助系统以保证受限时闭环系统的稳定性。分别采用动态逆和反演方法设计速度与高度子系统控制器,利用RBF神经网络逼近控制律的饱和特性,设计了一种非线性干扰观测器对模型不确定参数进行自适应估计,并在控制律中引入不依赖扰动上界的鲁棒项,对未观测的扰动部分进行自适应补偿,以保证控制律的强鲁棒性。引入跟踪微分器估计虚拟控制量的导数,解决了传统反演控制中"微分膨胀"问题。Lyapunov函数分析证明了闭环系统所有信号最终一致有界,闭环系统稳定。仿真结果表明:所提的控制策略能有效处理控制输入饱和问题,在受限情况下实现速度和高度对参考输入的高精度稳定跟踪,并对模型不确定性具较强的鲁棒性。

An adaptive back-stepping control approach based on RBF neural network compensation was proposed for hypersonic vehicles with input constraints and parametric uncertainties in this paper.The motion model in longitudinal direction of the vehicle was established.The constrains of fule-air ratio and elevation deflection angle were analyzed,which were caused by the actuator of the control system,elastic vibration and thermally choke avoiding in the chamber of the scramjet.The stability of the closed-loop system was guaranteed through design of an assistant system.The controllers of velocity subsystem and altitude subsystem were designed based on dynamic inversion and back-stepping respectively.RBF neural network was applied to approximate the saturation property of control laws.A nonlinear disturbance observer was designed to estimate the model uncertainties adaptively.The robust items that were independent of the upper bound of disturbance were introduced to compensate the unobserved disturbances,which guaranteed the developed controller's strong robustness.In order to solve the problem of differentiation explosion in the traditional back-stepping control,tracking differentiator was used for estimating the derivatives of virtual control laws.The analysis result of Lyapunov function approved that all signals in the closedloop system were consistent and limitary at last.The closed-loop system was stable.The simulation results show that the control strategy proposed can deal with the problem of control input constraints and realize the stable tracking velocity and height relative the reference input with high accuracy under the constrains.The control system has strong robustness to model uncertainties.