摘要
为解决一类不确定非线性系统的控制问题及系统混合干扰上界在实际应用中难以测量的问题,提出递归小脑神经网络模型分解控制算法。将系统分为名义模型、结构不确定性和非结构不确定性,分别对名义模型设计直接反馈控制器、对结构不确定性设计自适应控制器、对非结构不确定性设计鲁棒控制器。设计递归小脑模型关节控制器作为观测器来对系统干扰的上界进行实时逼近。李亚普诺夫理论证明了控制算法可使系统渐进稳定,微飞行机器人姿态控制仿真结果表明,控制算法改善了系统的动态性能及鲁棒性。研究结论为复杂非线性系统的有效控制提供了依据。
Model decomposition algorithm with recurrent cerebellar model articulation controller（CMAC） was proposed for a class of uncertain nonlinear systems whose upper boundary of lumped disturbance is difficult to measure in practice.The system was divided into nominal model,structured uncertainty,and unstructured uncertainty.A direct feedback controller was contrived for the nominal model,an adaptive controller was designed for the structured uncertainty,and a robust controller was schemed for the unstructured uncertainty respectively.The recurrent CMAC was framed as an observer to approximate the upper boundary of lumped disturbance in real-time.The asymptotically stability was proved based on Lyapunov＇s stability theory,and simulation results of micro flying robot attitude control indicated that the proposed algorithm improves transient performance and robustness.Research conclusions provide the basis for effective control of complex nonlinear systems.
出处
《电机与控制学报》
EI
CSCD
北大核心
2011年第1期91-97,共7页
Electric Machines and Control
基金
航天支撑技术基金(2007-HT-HGD-7)
关键词
非线性
自适应控制
鲁棒控制
小脑模型关节控制器
飞行机器人
nonlinearity
adaptive control
robust control
cerebellar model articulation controller
flying robot
