Robust H ∞ filtering for networked systems with multiple state delays
Robust H ∞ filtering for networked systems with multiple state delays
复制标题
DOI：
10.1080/00207170701196968
复制
发表时间：
2007-08
影响因子：
2.1
通讯作者：
Xiao He;Zidong Wang;Donghua Zhou
中科院分区：
文献类型：
--
作者：
Xiao He;Zidong Wang;Donghua Zhou
本文针对一类具有多状态延迟的网络系统，研究了一种新的鲁棒H ∞ 滤波器设计问题。同时考虑两种不完全测量，即具有随机延迟的测量和具有随机丢失现象的测量。这种不完整的测量是由通信网络的有限带宽引起的，并且被建模为依赖于相同随机变量的特定指示函数集的线性函数。重点关注全阶鲁棒 H ∞ 滤波器的分析和设计问题，使得对于所有允许的参数不确定性和所有可能的不完整测量，滤波误差动态呈指数均方稳定，并保证规定的 H ∞ 衰减水平。最近报道的一些方法，例如依赖于延迟和依赖于参数的稳定性分析方法，被用来获得不太保守的结果。根据某些线性矩阵不等式 (LMI)，为所需滤波器的存在建立了充分条件，这些条件取决于随机传感器延迟和丢失测量的发生概率。当这些 LMI 可行时，也可以表征所需滤波器的显式表达式。最后，通过数值算例说明了所提设计方法的有效性和适用性。
In this paper, a new robust H ∞ filter design problem is studied for a class of networked systems with multiple state-delays. Two kinds of incomplete measurements, namely, measurements with random delays and measurements with stochastic missing phenomenon, are simultaneously considered. Such incomplete measurements are induced by the limited bandwidth of communication networks, and are modelled as a linear function of a certain set of indicator functions that depend on the same stochastic variable. Attention is focused on the analysis and design problems of a full-order robust H ∞ filter such that, for all admissible parameter uncertainties and all possible incomplete measurements, the filtering error dynamics is exponentially mean-square stable and a prescribed H ∞ attenuation level is guaranteed. Some recently reported methodologies, such as delay-dependent and parameter-dependent stability analysis approaches, are employed to obtain less conservative results. Sufficient conditions, which are dependent on the occurrence probability of both the random sensor delay and missing measurement, are established for the existence of the desired filters in terms of certain linear matrix inequalities (LMIs). When these LMIs are feasible, the explicit expression of the desired filter can also be characterized. Finally, numerical examples are given to illustrate the effectiveness and applicability of the proposed design method.