|
Abstract: |
This paper proposes a state estimation method for a class of norm-bounded non-linear sampled-data descriptor systems using the Kalman filtering method. The descriptor model is firstly discretized to obtain a discrete-time non-singular one. Then a model of robust extended Kalman filter is proposed for the state estimation based on the discretized non-linear non-singular system. As parameters are introduced in for transforming descriptor systems into non-singular ones, there exist uncertainties in the state of the systems. To solve this problem, an optimized upper-bound is proposed so that the convergence of the estimation error co-variance matrix is guaranteed in the paper. A simulating example is proposed to verify the validity of this method at last. |
Key words: sampled-data system, descriptor system, state estimation, Kalman filtering, REKF |
DOI:10.11916/j.issn.1005-9113.17143 |
Clc Number:TP13 |
Fund: |
|
Descriptions in Chinese: |
非线性广义采样系统的状态估计—鲁棒扩展卡尔曼滤波方法 王茂1,梁天添1,周振华2 (1.哈尔滨工业大学 空间控制与惯性技术研究中心,哈尔滨 150001; 2.常州工业职业技术学院,江苏 常州 213000) 创新点说明:1)引入参数并使用欧拉离散化方法将广义连续-离散系统转化为非奇异一般系统进行研究; 2)针对转化得来的非奇异系统中存在状态新增不确定性的问题,提出鲁棒上界,并根据该上界设计鲁棒扩展卡尔曼滤波器(REKF); 3)对设计的滤波器鲁棒性进行证明,证明该滤波器符合鲁棒性能指标。 研究目的、方法: 为解决一类范数有界非线性广义连续-离散系统的状态估计问题,在Matlab环境下,对REKF与EKF算法进行仿真对比 研究结果: 仿真结果表明,由REKF算法得到的状态估计误差小于EKF算法得到结果的20%,相较于EKF算法,REKF算法能更好估计该类广义系统的状态。 结论: 1)提出的REKF算法可有效解决系统非线性问题; 2)提出的鲁棒算法可有效降低非奇异等价系统中新增状态不确定性对状态估计影响; 3)相较于EKF算法,本文提出的REKF算法能够保证状态估计误差协方差矩阵有效收敛,从而能更好的估计该类广义连续-离散系统的状态。 关键词:采样系统;广义系统;状态估计;卡尔曼滤波;REKF算法 |