Robust passive filtering for neutral-type neural networks with time-varying discrete and unbounded distributed delays
Robust passive filtering for neutral-type neural networks with time-varying discrete and unbounded distributed delays
复制标题
DOI：
10.1016/j.jfranklin.2013.01.021
复制
发表时间：
2013-06
期刊：
影响因子：
--
通讯作者：
X. Lin;Xian Zhang;Yantao Wang
中科院分区：
文献类型：
--
作者：
X. Lin;Xian Zhang;Yantao Wang
研究了一类具有时变离散、无界分布时滞的中性型神经网络的无源滤波问题。基于无源理论，给出了鲁棒无源滤波器存在的充分条件。通过引入适当的Lyapunov-Krasovskii泛函并利用Jensen不等式技术处理其导数，以非线性矩阵不等式的形式给出了耗散γ>0的误差动态系统严格无源的判据。为了解决非线性问题，提出了锥互补线性化（CCL）算法。此外，当神经网络类中出现范数有界参数不确定性时，还研究了相应的鲁棒被动滤波问题。给出了三个例子来证明该方法的有效性。
The passive filtering problem is studied for a class of neutral-type neural networks with time-varying discrete and unbounded distributed delays. Based on the passive theory, a sufficient condition for the existence of the robust passive filter is given. By introducing an appropriate Lyapunov–Krasovskii functional and using Jensen's inequality technique to deal with its derivative, the criterion which ensures error dynamic system to be strictly passive with dissipation γ>0 is presented in the form of nonlinear matrix inequality. In order to solve the nonlinear problem, a cone complementarity linearization (CCL) algorithm is proposed. Furthermore, when the norm-bounded parameter uncertainties appear in the class of neural networks, the corresponding robust passive filtering problem is also investigated. Three examples are given to demonstrate the effectiveness of the proposed method.