Namadchian Zahra, Rouhani Modjtaba
IEEE Trans Neural Netw Learn Syst. 2018 Dec;29(12):5859-5869. doi: 10.1109/TNNLS.2018.2815579. Epub 2018 Apr 5.
This paper aims to analyze the problem of adaptive neural network (NN) tracking control for a class of switched stochastic nonlinear pure-feedback systems with unknown direction hysteresis. In the light of recent studies on the hysteresis phenomenon in the field of nonlinear switched systems, this paper focuses on Bouc-Wen hysteresis model with unknown parameters and direction conditions. To simplify the control design, the following procedure is applied. Prior to tackling the unknown direction hysteresis problem based on the Nussbaum function and the backstepping techniques, the pure-feedback structure difficulty is governed by the mean value theorem. Furthermore, an optimized adaptation method is utilized to cope with computational burden. Universal approximation capability of radial basis function NNs and Lyapunov function method is synthesized to develop an adaptive NN tracking control scheme. It is demonstrated that under arbitrary deterministic switching, the presented controller can guarantee that all signals in the closed-loop system are semiglobally uniformly ultimately bounded in probability and the tracking error converges to a neighborhood of the origin. Finally, two simulation examples are given to illustrate the advantages of the proposed control design approach.
本文旨在分析一类具有未知方向滞后的切换随机非线性纯反馈系统的自适应神经网络（NN）跟踪控制问题。鉴于近期非线性切换系统领域中关于滞后现象的研究，本文聚焦于具有未知参数和方向条件的Bouc-Wen滞后模型。为简化控制设计，采用以下步骤。在基于Nussbaum函数和反步技术解决未知方向滞后问题之前，利用均值定理解决纯反馈结构难题。此外，采用一种优化的自适应方法来应对计算负担。综合径向基函数神经网络的通用逼近能力和李雅普诺夫函数方法，设计了一种自适应神经网络跟踪控制方案。结果表明，在任意确定性切换下，所提出的控制器能够保证闭环系统中的所有信号在概率意义下半全局一致最终有界，并且跟踪误差收敛到原点的一个邻域内。最后，给出两个仿真例子来说明所提出的控制设计方法的优点。
