几类具有未建模动态非线性系统自适应神经网络控制

发布时间：2018-05-29 23:33

本文选题：非线性系统 + 自适应神经网络控制　；参考：《渤海大学》2017年硕士论文

【摘要】：本文在国内外不确定非线性系统相关研究基础上,应用神经网络控制理论,结合自适应反步递推设计和鲁棒控制理论,研究了具有未建模动态非线性系统的控制问题,提出有效的自适应神经网络控制方法.本文主要从下面两个部分进行论述:1.针对一类具有未建模动态的非仿射非线性时滞系统,发展了一种自适应神经网络智能控制方案。在控制器的设计过程中,应用变量分离技术克服系统的全状态时滞函数的设计困难,利用反步递推设计方法和神经网络的万能逼近能力提出了能够保证闭环系统所有信号一致最终有界的自适应神经网络控制方案。仿真结果验证了所提出的控制方案的有效性。2.研究了一类具有未建模动态和输入饱和的严格反馈非线性系统的自适应神经网络控制问题。在控制设计过程中,应用径向基函数神经网络近似逼近未知非线性函数;反步递推设计方法来构造一种自适应神经网络控制方案。所提出的控制方案保证了闭环系统的半全局有界性,同时通过估计神经网络权向量范数的最大值,使得控制系统只需要一个自适应参数,从而减小了计算量.仿真结果验证了本章所提出的方法的有效性.
[Abstract]:In this paper, based on the research of uncertain nonlinear systems at home and abroad, the control problem of unmodeled dynamic nonlinear systems is studied by applying neural network control theory, combined with adaptive backstepping recursive design and robust control theory. An effective adaptive neural network control method is proposed. This paper mainly discusses the following two parts: 1. An adaptive neural network intelligent control scheme is developed for a class of non-affine nonlinear time-delay systems with unmodeled dynamics. In the design of the controller, the variable separation technique is used to overcome the difficulty of the design of the full state delay function of the system. Using the backstepping recursive design method and the universal approximation ability of the neural network, an adaptive neural network control scheme is proposed, which can guarantee the uniform and ultimately bounded signals of the closed-loop system. Simulation results verify the effectiveness of the proposed control scheme. The adaptive neural network control problem for a class of strictly feedback nonlinear systems with unmodeled dynamics and input saturation is studied. In the process of control design, the radial basis function neural network is used to approximate the unknown nonlinear function, and the backstepping recursive design method is used to construct an adaptive neural network control scheme. The proposed control scheme guarantees the semi-global boundedness of the closed-loop system, and by estimating the maximum value of the weight vector norm of the neural network, the control system only needs one adaptive parameter, thus reducing the computational complexity. The simulation results verify the effectiveness of the proposed method in this chapter.
【学位授予单位】：渤海大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP273;TP183

【参考文献】