基于递归小脑神经网络的模糊自适应控制

摘要

In order to solve the control problem for a class of uncertain nonlinear system,a fuzzy adaptive control algorithm for cerebellar neural network was proposed.The system was divided into the nominal model,parameter uncertainty section and hybrid interference item including the modeling error,disturbances and un-modeled dynamic.The fuzzy adaptive control was adopted to approach every uncertain parameter of the system in real time,and the robustness control was used to eliminate the hybrid interference.In addition,the recurrent cerebellar model articulation controller（CMAC） was designed as an observer to approximate the upper boundary of the hybrid interference in real time.The uniformly bounded stability of the system was proved based on Lyapunov＇s theory.The simulation results for the attitude control of micro flying robot indicate that the proposed control algorithm improves the dynamic performance and robustness of the system.And the research conclusions can provide the basis for the effective control of complex nonliear systems.%为解决一类不确定非线性系统控制问题,提出了小脑神经网络模糊自适应算法.将系统分为标称模型、参数不确定部分以及包含建模误差、干扰及未建模动态等在内的混合干扰项,用模糊自适应控制实时逼近系统各个不确定参数,用鲁棒控制消除混合干扰,并设计了递归小脑模型关节控制器作为观测器来对混合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一致有界稳定,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,研究结论对复杂非线性系统的有效控制提供了依据.