引用本文:周晗,孙书利.网络化随机不确定系统基于SCP的最优线性估值器[J].控制理论与应用,2025,42(7):1379~1387.[点击复制]
ZHOU Han,SUN Shu-li.Optimal linear estimators for networked stochastic uncertain systems under SCP[J].Control Theory & Applications,2025,42(7):1379~1387.[点击复制]
网络化随机不确定系统基于SCP的最优线性估值器
Optimal linear estimators for networked stochastic uncertain systems under SCP
摘要点击 3415  全文点击 206  投稿时间：2023-08-02  修订日期：2025-04-24
查看全文  查看/发表评论  下载PDF阅读器   HTML
DOI编号  10.7641/CTA.2024.30524
  2025,42(7):1379-1387
中文关键词  网络化系统  随机通信协议(SCP)  线性估值器  随机参数矩阵  相关噪声
英文关键词  networked system  stochastic communication protocol (SCP)  linear estimator  random parameter matrix  correlated noise
基金项目  国家自然科学基金项目(62473132), 黑龙江省自然科学基金重点项目(ZD2021F003)资助.
作者单位E-mail
周晗 黑龙江大学 电子工程学院 zhouhan1130@foxmail.com 
孙书利* 黑龙江大学 电子工程学院 sunsl@hlju.edu.cn 
中文摘要
      在随机通信协议(SCP)调度下, 针对带随机参数矩阵和相关噪声的网络化线性离散时变系统, 研究了依赖 SCP通信概率的最优状态估计问题. 采用白噪声描述系统的随机参数不确定性. 当传感器经过有限带宽的通信网络 传输观测数据给估值器时, 为了避免数据拥塞采用SCP对各观测分量进行调度, 保证每时刻只传输一个观测分量. 在估值器不知道收到的是传感器的哪一个观测分量而只知道各分量的通信概率的情况下, 应用射影理论设计了依 赖各观测分量传输概率的最优线性递推预报器、滤波器与平滑器. 问题归结为求解一个简单的差分方程, 一个Lyapunov方程和一个Riccati方程. 分析了算法的稳态特性, 给出了稳态存在的一个充分条件. 仿真结果验证了所提算法 的有效性.
英文摘要
      The optimal state estimation problem based on stochastic communication protocol (SCP) is studied for networked linear discrete time-varying systems with random parameter matrices and correlated noises, where the communication probabilities are governed by the SCP, white noises are used to characterize the uncertainties of stochastic parameters. When the sensor transmits measurement data to the estimator through a bandwidth-limited communication network, the SCP is used to schedule each measurement component of a sensor to avoid data congestion and ensure that only one measurement component is transmitted at each time. In scenarios where the estimator does not know which measurement component of the sensor is received but only knows the communication probability of each component, the optimal linear recursive predictor, filter, and smoother that depend on the transmission probability of each component are designed in the linear minimum variance sense by applying the projection theory. The solution involves solving a simple difference equation, a Lyapunov equation, and a Riccati equation. The steady-state property of the algorithms is analyzed, and a sufficient condition for the existence of the steady state is given. Simulation results demonstrate the effectiveness of the proposed algorithms.