三自由度直升机基于无模型自适应的姿态控制
Model free adaptive attitude control for a 3-DOF helicopter
王介港 1王向华 2王向荣 3张子叶 4王建东1
作者信息
- 1. 山东科技大学电气与自动化工程学院,山东青岛 266590
- 2. 北京邮电大学人工智能学院,北京 100876
- 3. 北京航空航天大学电子信息学院,北京 100191
- 4. 山东科技大学数学与系统科学学院,山东青岛 266590
- 折叠
摘要
本文针对三自由度直升机姿态控制系统改进了无模型自适应控制方法(MFAC).MFAC是数据驱动的控制方式,其只利用系统输入、输出数据,不依赖于系统数学模型.然而现有基于紧格式动态线性化数据模型的MFAC方法要求系统满足4个假设.本文首先证明了三自由度直升机姿态控制系统虽然满足假设1-3,但不满足假设4(控制输入增加时相应的受控系统输出应该是不减的).为此,本文针对三自由度直升机姿态控制系统改进了MFAC,具体有:1)定义新的输出为角度与角速度的线性组合,并证明了新定义的输出满足假设4;2)证明了新定义输出中所引入的线性参数影响了闭环系统性能;3)增加了输入微分项,并证明了输入微分的引入可改善闭环系统特征根的分布,提高系统性能.最后,本文通过仿真实验和实物实验分别验证了所提方案的有效性.
Abstract
In this paper,an improvement to model free adaptive control(MFAC)method is proposed for attitude control system of a 3-DOF helicopter.MFAC is a data-driven control method and generates the control signals based on the input and output data,not relying on the system mathematical model.The existing MFAC methods based on the compact form dynamic linearized data model require the system to satisfy four assumptions,however,this paper demonstrates that the 3-DOF helicopter attitude control system satisfies all other assumptions except for Assumption 4(the corresponding controlled system output should be non-decreasing when the control input increases).Hence for the attitude control system of a 3-DOF helicopter,following improvements are made to the existing MFAC methods:i)A new output is defined as a linear combination of angle and angular velocity,and it has been proved that the redefined output satisfies Assumption 4;ii)It is proved that the linear parameter in the redefined output influences the performance of closed-loop system;iii)The input differentiation term is added to change the eigenvalue distribution of the closed-loop system and then improve the system performance.Finally,simulations as well as experiments are conducted to verify the effectiveness of the proposed scheme,respectively.
关键词
无模型自适应控制/三自由度直升机/数据驱动控制/姿态控制Key words
model free adaptive control/3-DOF helicopter/data-driven control/attitude control引用本文复制引用
出版年
2024
国家科技期刊平台
NSTL
万方数据