带未知观测输入随机不确定系统的状态估计

发布时间：2018-06-21 08:41

本文选题：未知观测输入 + 观测丢失　；参考：《黑龙江大学》2015年硕士论文

【摘要】：含未知输入干扰随机系统的状态估计问题广泛存在于控制、信号处理和故障诊断等应用中。在许多情况下,外界扰动往往是无法测量的(即未知输入),如果不能很好地对干扰或故障进行有效的检测和分离,可能会造成人员和生产上的损失。此外,在网络化控制系统(NCSs)中,由于网络的带宽和承载能力有限,数据在传输时发生随机时滞和观测丢失的现象不可避免。除了以上干扰,由乘性噪声描述的参数不确定性也可能存在于系统的模型中。以往的文献大都分别对带有未知输入、时滞、观测丢失和乘性噪声的系统开展研究,但同时考虑以上不确定性的文献鲜见。因此,考虑到这些问题,本文研究带未知观测输入随机不确定网络化系统的状态融合估计问题。主要研究内容如下:对同时具有未知观测输入、观测数据丢失和参数乘性噪声的随机不确定系统,提出了与未知观测输入解耦的具有Kalman形式的分布式和集中式融合滤波器,包括先验滤波器(一步预报器)和后验滤波器。给出了任意两个传感器子系统之间的滤波误差互协方差阵。对于相应的定常系统,分别给出了分布式和集中式融合稳态滤波器存在的充分条件,并证明了任意两个传感器子系统之间的互协方差阵稳态解的存在性。最后,给出了未知输入的次优估计算法。对同时具有未知观测输入和一步随机时滞的随机不确定系统,通过引入新的变量,将带未知观测输入和随机时滞的系统转化为等价的带随机参数的系统,提出了与未知观测输入解耦的分布式和集中式融合滤波器,包括先验滤波器(一步预报器)和后验滤波器。给出了任意两个传感器子系统之间的滤波误差互协方差阵。最后,给出了未知输入的次优估计算法。
[Abstract]:The problem of state estimation for stochastic systems with unknown input disturbances is widely used in control, signal processing and fault diagnosis. In many cases, external disturbances are often unmeasurable (I. e., unknown inputs). If interference or fault can not be effectively detected and separated, it may result in loss of personnel and production. In addition, in networked control system (NCSs), due to the limited bandwidth and carrying capacity of the network, the phenomenon of random delay and observation loss in data transmission is inevitable. In addition to the above disturbances, the parametric uncertainties described by multiplicative noise may also exist in the system model. Most of the previous literatures have studied the systems with unknown input, time delay, observation loss and multiplicative noise, but there are few papers considering the uncertainties mentioned above. Therefore, considering these problems, the problem of state fusion estimation for networked systems with unknown observation inputs is studied in this paper. The main research contents are as follows: for stochastic uncertain systems with unknown input, data loss and parametric multiplicative noise, a distributed and centralized fusion filter with Kalman form is proposed, which is decoupled from unknown observation input. A priori filter (one-step predictor) and a posteriori filter are included. The filter error mutual covariance matrix between any two sensor subsystems is given. For the corresponding time-invariant systems, sufficient conditions for the existence of distributed and centralized fusion steady-state filters are given, and the existence of the steady-state solutions of the cross-covariance matrix between any two sensor subsystems is proved. Finally, a suboptimal estimation algorithm for unknown inputs is presented. For stochastic uncertain systems with both unknown observation input and one step stochastic delay, the system with unknown observation input and stochastic delay is transformed into an equivalent system with random parameters by introducing new variables. A distributed and centralized fusion filter, which is decoupled from unknown observation inputs, is proposed, including a priori filter (a one-step predictor) and a posteriori filter. The filter error mutual covariance matrix between any two sensor subsystems is given. Finally, a suboptimal estimation algorithm for unknown inputs is presented.
【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】：TN713

【相似文献】