基于神经网络的时滞广义系统鲁棒H_∞控制  

Robust H_∞ Control for Time-delay Singular Systems Based on Neural Networks

作　　者：杨帆[1] 张庆灵[1] 

机构地区：[1]东北大学理学院

出　　处：《控制工程》2006年第4期324-326,373,共4页Control Engineering of China

基　　金：辽宁省普通高校学科带头人基金资助项目(124210);辽宁省科技基金资助项目(2001401041)

摘　　要：针对神经网络模型自身的结构特点,研究了当系统存在扰动时的鲁棒H∞控制问题。将神经网络和控制理论相结合,利用线性矩阵不等式方法和神经网络良好的逼近能力,对一类时滞广义系统设计了鲁棒H∞控制器,并给出了H∞反馈控制器存在的充分条件。使得在一定条件下,闭环系统渐近稳定且从扰动到控制输出的传递函数的H∞范数不超过某个确定的上界。所得结论可化为标准的线性矩阵不等式(LMI)求解,该方法具有一定的实际意义和可行性,算例也验证了其有效性。According to the composition of the mathematics model, the problem of robust H∞ control is considered. Connecting the neural network with H∞ control theory for a class of time-delay singular systems in connection with linear matrix inequality and the approximate ability of neural network, the robust H∞ controller is designed. Under some condition, the closed-loop system is asymptotically stable and the norm of the transfer function from exogenous disturbance to output is not more than a prescribed upper bound. Sufficient conditions for the existence of such： controller are derived based on the linear matrix inequality. Furthermore, it is shown that the result can be constructed in terms of the feasible solutions to the certain linear-matrix inequalities. Therefore, it has some real significance and feasibility. An example shows the effectiveness of the design method.

关 键 词：不确定广义系统 时滞 鲁棒H∞控制 线性矩阵不等式 

分 类 号：TP13[自动化与计算机技术—控制理论与控制工程]