Network-based H∞ state estimation for neural networks using imperfect measurement
Network-based H∞ state estimation for neural networks using imperfect measurement
复制标题
DOI：
10.1016/j.amc.2017.08.034
复制
发表时间：
2018
期刊：
影响因子：
--
通讯作者：
Tae H. Lee;Ju H. Park;Ho-Youl Jung
中科院分区：
文献类型：
--
作者：
Tae H. Lee;Ju H. Park;Ho-Youl Jung
研究了在采样效应、外部干扰、网络诱导时延和丢包等网络约束条件下神经网络的H∞状态估计问题。由于采样效应，外部干扰、网络引起的延迟和数据包丢失仅在采样时刻影响测量。此外，当发生数据包丢失时，使用最后接收的数据。针对非理想信号，设计了一种补偿器，并利用该补偿器设计了保证期望性能的H∞滤波器。数值算例验证了所提方法的有效性。
This study considers the network-based H∞ state estimation problem for neural networks where transmitted measurements suffer from the sampling effect, external disturbance, network-induced delay, and packet dropout as network constraints. The external disturbance, network-induced delay, and packet dropout affect the measurements at only the sampling instants owing to the sampling effect. In addition, when packet dropout occurs, the last received data are used. To tackle the imperfect signals, a compensator is designed, and then by aid of the compensator, H∞ filter which guarantees desired performance is designed as well. A numerical example is given to illustrate the validity of the proposed methods.