基于逼近的动态面滑膜智能控制算法的研究  

Research on Dynamic Surface Sliding Mode Intelligent Control Algorithm Based on Approximation

作　　者：邹晶[1] 李雷[1] 

机构地区：[1]南京邮电大学自动化学院,江苏南京210023

出　　处：《计算机技术与发展》2015年第6期6-11,共6页Computer Technology and Development

基　　金：国家自然科学基金资助项目(61373137;61070234;61071167)

摘　　要：文中研究的是自适应模糊神经网络(FNN)在一类带有未知延时函数和不确定干扰的非仿射纯反馈系统中的控制问题。首先采用动态表面控制(DSC)技术避免反演设计存在的"微分爆炸"问题,同时为了克服干扰,保证鲁棒性,引入滑模控制来改进DSC方法。然后直接利用模糊神经网络近似未知函数,并利用Lyapunov-Krasovskii泛函,双曲正切函数的特性以及函数分离技术去克服未知时间延迟函数问题。该算法只需少量的自适应参数,就能保证闭环系统中的所有信号半全局一致最终有界。仿真结果表明,在存在干扰和时滞的情况下,该控制器具有良好的实时性和稳定性。In this paper,consider the problem of adaptive Fuzzy Neural Network （ FNN） control for a class of nonaffine pure-feedback systems with unknown time-delay functions and perturbed uncertainties. Dynamic Surface Control （ DSC） technique is used to avoid the“explosion of complexity” problem. To overcome the interference and ensure the robustness, modify the DSC approach by employing sliding mode control. The FNN is directly utilized to approximate unknown functions,and use the Lyapunov-Krasovskii functions,the de-sirable property of hyperbolic tangent functions,and the function separation technique to overcome the problem from unknown time-delay functions. The proposed algorithm just needing a small number of adaptive parameters,can guarantee all the signals in the closed-loop system to be semiglobally uniformly ultimately bounded. The simulation results illustrate that the controller has good real-time and stability under the case of interference and time-delay.

关 键 词：自适应控制 反推设计 动态面滑模控制 模糊神经网络 非线性时滞系统 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]