Abstract

The paper deals with robust filtering algorithms for discrete systems with unknown inputs (disturbances) and Markovian jump parameter. The proposed filtering algorithm is based on the separation principle, minimization of a quadratic criterion and the use of Kalman filters with unknown input and smoothing procedures. Solving a non-stationary problem is represented solving a two-point boundary value problem in kind of difference matrix equations. In the stationary case problem is represented matrix algebraic equations. Robustness ensures the stability of the filter dynamics when errors occur in identifying the jump parameter. An example is provided to illustrate the proposed approach, which showed that the use of smoothing procedures for estimating an unknown input improves the accuracy of estimates.

中文翻译：

---

具有未知输入和跳转参数的离散系统的鲁棒滤波

摘要

本文针对具有未知输入（干扰）和马尔可夫跳跃参数的离散系统提出了鲁棒的滤波算法。所提出的滤波算法基于分离原理，最小化二次准则以及使用未知输入和平滑过程的卡尔曼滤波器。解决了一个非平稳问题，代表了一种用差分矩阵方程求解两点边值问题。在平稳情况下，问题用矩阵代数方程表示。鲁棒性可确保在识别跳跃参数时发生错误时滤波器动态的稳定性。提供一个示例来说明所提出的方法，该方法表明使用平滑过程估计未知输入可提高估计的准确性。