改进模型参考自适应控制及其在解耦控制中的应用

Improved model reference adaptive control and its application in decoupled control

提出了一种基于扩张状态观测器(extended state observer,ESO)的任意参考模型自适应控制方法,解决了被控对象状态信息不可测以及存在不确定因素导致模型参考自适应控制(model reference adaptive control,MRAC)效果变差甚至使系统不稳定的问题。在简述ESO教学模型的基础上提出了基于ESO的模型参考自适应控制方法并进行了严格的稳定性分析,仿真结果表明所设计方法具有跟踪速度快、稳态误差小、控制量小以及参考模型容易选择等特点。然后,提出基于改进MRAC的解耦控制方法并应用于三输入三输出的四旋翼飞行器姿态控制系统解耦控制,仿真结果表明该解耦控制方法具有鲁棒性好的特点。该方法无需按照通常的做法设计神经网络对不确定因素导致的误差进行补偿,大大简化了控制器设计过程。

An improved adaptive control with arbitrary reference mode was presented based on the extend-ed state observer （ ESO） , focusing on solving the problems：（ i） the system state is not measurable;（ ii） the model reference adaptive control （ MRAC ） is less effective or even make the system unstable when there exist internal or external uncertainties. After briefly describing the mathematical model of ESO, the improved MRAC was put forward and a strict stability analysis was given. The simulation results show that the improved MRAC possesses quick tracking rate, high steady state accuracy and small control moment, including easily choosing a reference model. Then, a robust decoupling control method based on the im-proved MRAC was proposed and applied to decoupling control of the attitude control for the quadrotor air-craft with three inputs and three outputs. The simulation result shows that it has strong robustness. It is not necessary to design a neural network as usual to compensate for the uncertainties, which greatly sim-plified the process of controller design.