集成神经网络与自适应算法的分数阶滑模控制

针对被控对象的参数时变和外部扰动问题,本文融合神经网络的万能逼近能力和自适应控制技术,并结合分数阶微积分理论,提出了基于神经网络和自适应控制算法的分数阶滑模控制策略.本文采用等效控制的方法设计滑模控制律,并利用神经网络的万能逼近能力估测控制律的变化,结合自适应控制算法和分数阶微积分理论抑制传统滑模控制系统的抖震,同时根据Lyapunov稳定性理论分析了系统的稳定性,最后给出了实验结果.实验结果表明,本文提出的基于神经网络和自适应控制算法的分数阶滑模控制系统,能保持滑模控制器对系统外部扰动和参数变化鲁棒性的同时,也能有效地抑制抖震,使得系统获得较高的控制性能.

Neural network and adaptive algorithm-based fractional order sliding mode controller

In this paper, a fractional order sliding mode scheme based on neural network self-adapting algorithm is proposed for dealing with the chattering phenomenon existing in conventional sliding mode controller under the existence of parameters variation and external disturbance. Firstly, the fractional order sliding mode control law is designed using equivalent control technology. And a switching control method is obtained to drive the system state to reach the given sliding manifold at any initial condition. Then the neural network and adaptive control algorithm are designed to abate the chattering of sliding mode controller. The stability of control system is analysis by Lyapunov stability theory finally. Experiments demonstrate that the proposed neural network self-adapting based fractional order sliding mode controller not only achieve better control performance than the conventional sliding mode control system, but also is robust with regard to system parameters variation and external disturbance.