多步随机滞后和多丢包网络系统的融合滤波

发布时间：2018-03-07 22:12

本文选题：随机时滞　切入点：丢包　出处：《黑龙江大学》2015年硕士论文　论文类型：学位论文

【摘要】：随着网络技术、通信技术和自动控制技术的融合与发展,网络化控制系统已成为人们关注的焦点。网络控制系统因具有成本低、易资源共享和远程操作等优点,而被广泛用于国民经济建设中。但随着网络系统规模的增大,复杂程度的增加,因网络承载能力和通信带宽的有限性使得网络数据交换往往存在随机滞后和丢包现象。此外,环境干扰以及传感器自身损耗也会使系统具有不确定性。针对上述问题,本文应用射影理论和线性最小方差意义下的分布式融合算法,研究具有多随机时滞和丢包网络化随机不确定系统的融合滤波问题,主要研究内容如下:对数据包带有和不带有时间戳的随机滞后和丢包的多传感器离散线性随机系统,分别推导了依赖随机变量值和依赖随机变量概率的任两个传感器子系统之间的滤波误差互协方差阵。基于已有的局部滤波器和所推得的滤波误差互协方差阵,分别设计了依赖随机变量值和依赖随机变量概率的分布式和集中式融合滤波器,进一步,对依赖随机变量概率的分布式和集中式融合滤波器,分析了稳态特性,并给出了稳态融合滤波器存在的一个充分条件。对带有随机乘性噪声和多步随机滞后多丢包的多传感器离散线性随机系统,提出了局部子系统的线性最小方差最优线性滤波器。当状态方程和观测方程中均含有乘性噪声时,推导了任两个传感器子系统间的滤波误差互协方差阵,提出了分布式按矩阵加权融合滤波器和集中式融合滤波器,并给出了稳态融合滤波器存在的一个充分条件。进一步,当仅状态方程含有乘性噪声时,为了减小计算负担,又提出了避免计算互协方差阵的协方差交叉融合滤波器。对带有不确定观测、多随机滞后和丢包的多传感器线性离散随机系统,设计了线性最小方差意义下的局部最优线性滤波器,推导了任两个传感器子系统间的误差互协方差阵,给出了按矩阵加权分布式融合滤波器。对系统仅含有一步随机滞后情况,推导了集中式融合滤波器,并给出了稳态存在的一个充分条件。
[Abstract]:With the integration and development of network technology, communication technology and automatic control technology, networked control system has become the focus of attention. Network control system has the advantages of low cost, easy resource sharing and remote operation. But with the increase of the scale of network system and the increase of complexity, the network data exchange often has the phenomenon of random delay and packet loss due to the limitation of network carrying capacity and communication bandwidth. Environmental interference and sensor loss also make the system uncertain. In view of the above problems, this paper applies projective theory and distributed fusion algorithm in the sense of linear minimum variance. In this paper, the problem of fusion filtering for networked stochastic uncertain systems with multiple stochastic delays and packet loss is studied. The main contents are as follows: for multisensor discrete linear stochastic systems with and without timestamp, stochastic delay and packet loss are studied. The filtering error mutual covariance matrix between any two sensor subsystems depending on random variable value and probability of random variable is derived, respectively. Based on the existing local filter and the derived filtering error cross covariance matrix, Distributed and centralized fusion filters depending on the probability of random variables and random variables are designed, respectively. Furthermore, the steady-state characteristics of distributed and centralized fusion filters dependent on probability of random variables are analyzed. A sufficient condition for the existence of a steady-state fusion filter is given. For a multisensor discrete linear stochastic system with stochastic multiplicative noise and multi-step stochastic delay and multiple packet loss, A linear minimum variance optimal linear filter for local subsystems is proposed. When multiplicative noise is present in both the state equation and the observation equation, the filtering error covariance matrix between any two sensor subsystems is derived. A distributed matrix weighted fusion filter and a centralized fusion filter are proposed, and a sufficient condition for the existence of the steady-state fusion filter is given. Furthermore, when only the equation of state contains multiplicative noise, in order to reduce the computational burden, A covariance crossover fusion filter is proposed to avoid computing the cross covariance matrix. For multisensor linear discrete stochastic systems with uncertain observations, multiple stochastic delays and packet loss, the covariance cross fusion filter is proposed. A local optimal linear filter is designed in the sense of linear minimum variance. The error mutual covariance matrix between any two sensor subsystems is derived, and the distributed fusion filter weighted by matrix is given. A centralized fusion filter is derived and a sufficient condition for the existence of steady state is given.
【学位授予单位】：黑龙江大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP202;TN713

【相似文献】