项目名称: 控制方向未知的随机非线性系统的神经网络自适应控制
项目编号: No.61304071
项目类型: 青年科学基金项目
立项/批准年度: 2014
项目学科: 自动化技术、计算机技术
项目作者: 余昭旭
作者单位: 华东理工大学
项目金额: 24万元
中文摘要: 由于考虑了外部随机干扰、未知控制方向和系统不确定性等因素的影响,控制方向未知的随机非线性系统在实际工业过程中适用面更广,因此针对该类系统的控制设计与分析具有重要的理论与实际意义。 本项目将考虑多类控制方向未知的随机非线性系统的神经网络自适应控制问题。控制设计的主要技术困难来自于: 1)由伊藤随机微分引入的梯度项和高阶海塞矩阵项; 2)由控制增益函数符号未知而带来控制方向未知。通过探索Nussbaum增益函数的新特性及新应用,解决由控制方向未知带来的困难。对系统非线性函数进行合理假设,拟采用线性状态变换、Backstepping、动态面控制、神经网络逼近及参数化和输入-驱动观测器设计等方法,研究随机非线性系统的神经网络自适应控制器设计,通过理论分析与仿真实验验证控制设计的有效性。该项目的成功实施,将对机器人控制、化工过程控制及飞行器控制等受随机因素影响的非线性系统的控制性能改善产生积极影响。
中文关键词: 随机非线性系统;未知控制方向;自适应控制;神经网络;Nussbaum增益函数
英文摘要: Considering the affection of some factors, such as external random disturbance, unknown control direction and the uncertainty of system, the stocastic nonlinear system with unknown control directions is widely used in many industrial processes. Thus, control design and performance analysis for such system is of importance in theory and practice. This project focuses on the adaptive neural control for some classes of stochastic nonlinear systems with unknown control directions. Major technical difficulties for these classes of systems lie in: 1) the gradient and the higher order Hessian term involved by the Ito stochastic differentiation; 2) the unknown control directions from the unknown signs of control gain functions. By exploring the new properties and new applications of Nussbaum function to stochastic nonlinear systems, control directions can be dealt with. Moreover, based on a novel combination of linear state transformation,Backstepping technique,Dynamic Surface Control(DSC) method,Neural netwroks approximation and its parameterization technique,and Input-driven observer design,et al, some adaptive neural control schemes will be developed for such systems under some suitable assumptions on nonlinear system functions. Meanwhile,the effectiveness of control design will be verified by theoretic analysise
英文关键词: stochastic nonlinear system;unknown control direction;adaptive control;Neural network;Nussbaum gain function