具有输入约束的不确定非线性系统自适应神经网络控制

发布时间：2018-06-05 19:05

本文选题：非线性系统 + 随机系统　；参考：《南京理工大学》2016年博士论文

【摘要】：在现实世界中,大部分控制系统具有本质非线性、不确定性,而且通常受到输入约束等因素的影响.因此,具有输入约束的不确定非线性系统的研究引起了广泛的关注.但此类系统尚且没有统一的理论与方法,必须针对不同类型的输入约束和控制问题分别进行分析与处理.本文在已有工作的基础上,基于神经网络逼近理论,研究了具有输入约束的两类控制问题:一类是随机非线性系统的自适应控制;一类是非线性多智能体系统的分布式协调控制.主要研究工作如下:1.针对具有不同类型输入约束的不确定随机非线性时滞系统,研究了自适应神经网络有界镇定问题.首先,对于饱和输入,给出了单输入单输出的随机非线性时滞系统的自适应神经网络控制方案.其次,对于死区输入,设计了随机非线性时滞互联大系统自适应神经网络分散控制器.所提出的控制方案均保证了闭环系统的所有信号是依概率有界的.仿真算例表明了设计方案的可行性.2.针对带有输入约束的不确定切换随机非线性系统,研究了自适应神经网络跟踪控制问题.首先,考虑具有饱和输入的严格反馈切换随机非线性系统.结合神经网络逼近理论、反步递推设计方法和公共李雅普诺夫函数方法,给出了自适应神经网络跟踪控制方案.所提出的控制方案保证了闭环系统的所有信号是依概率有界的,而且跟踪误差可以收敛到原点的充分小邻域内.其次,考虑具有滞回特性的纯反馈切换随机非线性系统.利用均值定理将系统的非仿射函数转化为仿射形式,进而提出了基于神经网络逼近的自适应跟踪控制方案.数值算例说明了设计方法的有效性.3.针对具有饱和输入的非线性多智能体系统,研究了分布式一致性控制问题.首先,应用神经网络逼近未知非线性函数,引入辅助系统补偿饱和输入的影响并结合命令滤波技术和反步递推设计方法,获得了分布式一致性控制方案.然后利用李雅普诺夫稳定性理论,证明了闭环系统的所有信号是半全局一致最终有界的,而且一致性误差收敛到原点的小邻域内.在此基础上,进一步研究了一类带有执行器故障的非线性多智能体系统的分布式包含控制问题,提出了分布式包含控制器的有效设计方法.仿真结果表明了所提出控制方法的有效性.4.针对带有未知死区输入的非仿射非线性多智能体系统,研究了指定性能分布式一致性控制问题.首先应用神经网络逼近未知非线性函数并利用指定性能函数实现同步误差的稳态和暂态性能,然后结合李雅普诺夫稳定性理论和动态面控制技术,给出了不依赖于死区参数的分布式一致性控制方案.所提出的控制方案保证了闭环系统的所有信号是半全局一致最终有界的,而且跟随者的输出以预定义的误差同步于领导者信号.数值算例说明了控制方案的可行性.
[Abstract]:In the real world, most of the control systems have essential nonlinearity, uncertainty, and are usually influenced by factors such as input constraints. Therefore, the study of uncertain nonlinear systems with input constraints has aroused widespread concern. However, there is no unified theory and method for such systems, which must be based on different types of input. On the basis of the existing work, based on the neural network approximation theory, this paper studies two kinds of control problems with input constraints: one is the adaptive control of stochastic nonlinear systems and the other is the distributed coordinated control of nonlinear multi-agent systems. The main research work is as follows: 1. For uncertain stochastic nonlinear time-delay systems with different types of input constraints, an adaptive neural network bounded stabilization problem is studied. Firstly, for saturated input, an adaptive neural network control scheme for stochastic nonlinear time-delay systems with single input and single output is given. The proposed control scheme guarantees that all the signals of the closed loop system are bounded by probability. The simulation example shows the feasibility of the design scheme.2. for an uncertain switched stochastic nonlinear system with input constraints, and the adaptive neural network tracking control question is studied. First, a strict feedback switched stochastic nonlinear system with saturated input is considered. Combining the neural network approximation theory, the backstepping design method and the common Lyapunov function method, the adaptive neural network tracking control scheme is given. The proposed control scheme guarantees that all the signals of the closed loop system are bounded by probability. And the tracking error can be converged to the sufficient small neighborhood of the origin. Secondly, a pure feedback switched stochastic nonlinear system with hysteresis characteristics is considered. By means of the mean theorem, the non affine function of the system is transformed into an affine form, and a self-adaptive tracking control scheme based on neural network approximation is proposed. The effectiveness of the method.3. studies the distributed consistency control problem for the nonlinear multi-agent systems with saturated input. First, the neural network is applied to the unknown nonlinear function, the auxiliary system is introduced to compensate the influence of the saturated input, and the distributed consistency is obtained by combining the command filtering technique and the backstepping design method. By using Lyapunov stability theory, it is proved that all the signals in the closed loop system are semi global and final bounded, and the consistency error converges to the small neighborhood of the origin. On this basis, the distributed inclusion control problem of a class of non linear multi-agent systems with actuator failures is further studied. An effective design method of distributed inclusion controller is proposed. The simulation results show that the effectiveness of the proposed control method.4. is used to study the distributed conformance control problem of the designated performance for non affine nonlinear multi-agent systems with unknown dead zone input. The qualitative function realizes the steady-state and transient performance of the synchronization error. Then, combining the Lyapunov stability theory and the dynamic surface control technology, the distributed conformance control scheme is given without the dead zone parameters. The proposed control scheme guarantees that all the signal numbers of the closed loop system are semi global and final bounded, and the following are followed by the control scheme. The output is synchronized to the leader signal with predefined error. Numerical examples illustrate the feasibility of the control scheme.
【学位授予单位】：南京理工大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TP13;TP183

【参考文献】