基于AC学习的非奇异快速终端滑模控制  

Non-singular Fast Terminal Sliding Mode Control Based on Actor-Critic Learning

作　　者：周迎春[1] 刘伟[2] 黄辉先[2] 胡志军[1] ZHOU Ying-chun;LIU Wei;HUANG Hui-xian;HU Zhi-jun(College of Intelligent Manufacturing,Hunan Vocational Institute of Technology,Xiangtan 411104,China;College of Information Engineering,Xiangtan University,Xiangtan 411105,China)

机构地区：[1]湖南理工职业技术学院智能制造学院,湖南湘潭411104 [2]湘潭大学信息工程学院,湖南湘潭411105

出　　处：《控制工程》2021年第10期2045-2051,共7页Control Engineering of China

基　　金：湖南省教育厅一般项目(17C0742)。

摘　　要：针对一类SISO不确定非线性系统,提出一种基于执行器-评价器学习的非奇异快速终端滑模(ACNFTSM)控制方法。在保证控制器非奇异的情况下,设计一种非奇异快速终端滑模控制器,并通过理论分析证明了滑模面到达平衡点的时间小于传统非奇异终端滑模(NTSM)控制的到达时间。为了消除系统扰动量对控制性能的影响,采用执行器-评价器学习算法对控制器参数进行实时优化,增强系统鲁棒性。模拟仿真和实物实验的结果验证了该方法的有效性。For a class of SISO uncertain nonlinear systems,a non-singular fast terminal sliding mode control method based on actor-critic learning(ACNFTSM)is proposed.A non-singular fast terminal sliding mode controller is designed to guarantee the non-singularity of the controller.The theoretical analysis proves that the time for the sliding mode surface to reach the equilibrium point is less than that of the traditional non-singular terminal sliding mode(NTSM)control.In order to eliminate the influence of system disturbance on control performance,actor-critic learning algorithm is used to optimize the parameters of the controller in real time and enhance the robustness of the system.The simulation and experimental results verify the effectiveness of the method.

关 键 词：终端滑模 奇异问题 RBF神经网络 AC学习 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]