T-S模糊模型与实际系统有非线性差异时自适应控制

In this proposal, according to the differences of the existing T-S fuzzy models and its reality systems,the general modeling errors are approximated by unknown time-varying parametric neural networks,a novel T-S fuzzy model with general modeling errors is proposed.A new control strategy, which consists of a parallel distributed compensation feedback term and difference-differential-type adaptive compensated term, is proposed for the T-S fuzzy model with the general modeling errors. The sufficient condition of the stability for resulting closed-loop system is also given via LMIs and adaptive robust control theory. A novel multiple T-S fuzzy model with modeling errors and Markov switching Ito stochastic time-delay equations is presented for nonlinear stochastic time-delay systems, a new adaptive control strategy is designed. A sufficient condition of stochastic stability in term of bilinear matrix inequatity for the closed-loop systems is given by using stochastic piecewise quadratic Lyapunov-krasovksii function and weight free matrix and furthermore, a class of solving BMIs problem are changed into an LMI terms,so as to reduce greatly the difficulty grade.The condition of the finite-time stochastic stability based on the novel models with the general modeling errors is also proposed and the finite-time stochastic adaptive stablization strategy is established.The obtained results are then extended to the problems of synchronization of the complex stochastic time-delay dynamic networks, proposes a distributed adaptive synchronization algorithm. The proposed approaches will give a new way to solve the control problem of many real systems.

针对T-S模型和实际系统之间存在的一般差异，拟提出含有一般建模误差的新模糊T-S模型，用未知变参数化神经网路逼近建模误差，给出由并行分布补偿反馈项和差分-微分型自适应补偿项组成的控制器，利用LMI和鲁棒自适应控制理论，给出系统全局稳定的充分条件；对非线性随机时滞系统，建立带有一般建模误差和马尔可夫切换的多伊藤随机时滞方程的新T-S模糊模型，给出其随机稳定性的条件，设计系统的新鲁棒自适应控制器，利用随机Lyapunov稳定理论给出闭环系统稳定的新双线性矩阵不等式充分条件，并利用新的矩阵分解将一类双线性矩阵不等式转化为线性矩阵不等式，降低求解该问题的难度；给出基于该新模型的系统有限时间随机稳定与收敛的条件及其有限时间随机稳定化的条件；将以上结果推广到随机时滞复杂动态网同步问题，给出分布式自适应同步算法。为实际系统控制问题提供新途径。

本课题聚焦T-S模糊模型与实际系统存在差异问题， 分别研究单个复杂非线性系统的鲁棒与自适应模糊控制机理和网络化多个系统的鲁棒与自适应协同控制机理，取得的主要成果如下：对复杂非线性时滞系统，提出了一系列新的T-S模糊模型，给出了其模糊控制器和滤波器的设计方法，并证明了系统的鲁棒渐近稳定性；对复杂随机非线性系统，建立了几类新的T-S模型，设计了其鲁棒或自适应模糊控制器或滤波器，给出了系统随机渐近稳定、随机有限时间稳定的充分条件；对不确定网络化的非线性多智能体系统，基于T-S模糊系统模型，设计了几类分布式鲁棒自适应控制协议和鲁棒自适应学习控制协议，分别得到多智能体系统的渐近一致性和精确渐近一致性；对未知非线性多智能体系统，设计了一种全分布式自适应模糊控制协议，获得了非线性未知多智能体系统的全局一致性结果，克服了现有文献中半全局一致性的缺点。实现了具有无向连通图的二阶未知非线性多智能体系统的全局编队控制；对复杂动态网络系统，分别提出了自适应同步、自适应事件驱动拟周期间歇同步和事件驱动预设性能同步控制算法，并证明了相应网络系统的全局指数同步、渐近同步性，并改善了网络同步性能。.本项目共发表期刊论文43篇，书籍章节1部，会议论文16篇，其中SCI检索41篇，中科院大类分区1区4篇，2区15篇。取得的成果在T-S模糊模型与实际系统存在差异时，解决了几类复杂随机、时滞非线性系统和不确定网络化系统的控制和滤波问题。这些结果在网络化机器人系统、无人机群编队控制、信息物理系统和深空探测以及生物群体系统中有重要的应用价值。