一类非线性系统的自组织模糊神经网络控制

Control of a class of nonlinear systems based on self-organizing fuzzy neural

针对一类MIMO不确定非线性有干扰且控制增益符号未知的系统进行跟踪控制的问题,提出了一种在线自组织模糊神经网络的改进算法,用以克服参数选择困难的问题,并基于该算法给出了一种自适应鲁棒控制方法。首先基于主导输入的概念将MIMO系统分解为多个SISO系统构成的系统,然后结合自组织模糊神经网络在线对系统中的未知函数进行逼近,对网络结构和参数实现在线调节,再利用Nussbaum函数来克服控制增益符号未知,并且引入鲁棒项及复合误差的估计来补偿复合误差。最后基于Lyapunov稳定性理论证明了整个闭环系统半全局一致最终有界。理论和仿真结果表明提出方法的有效性。

This paper discusses the problem of tracking control a class of uncertain MIMO nonlinear sys- tem with disturbances and unknown control gain sign. An improved algorithm for online self-organizing fuzzy neural network is proposed to overcome the difficulty of choosing parameters and based on it, a ro- bust and adaptive controller is proposed. Firstly, the MIMO system was decompounded into several SISO systems, and then the improved self-organizing adaptive fuzzy neural network was utilized to approximate the unknown function with the structure and parameters being tuned online. Then a Nussbaum function was adopted to overcome the difficulty of the unknown control gain sign, and a robust control term and an error estimation term were utilized to compensate for the errors. The closed-loop control system was proved to be semi-globally, uniformly, and ultimately bounded by Lyapunov stability theorem. Theoretical analysis and numerical simulations show the effectiveness of the developed approach.