| 两级交互式多模型算法 |
| |
| 引用本文： | 梁彦,谭伟,潘泉,张洪才.两级交互式多模型算法[J].自动化学报,2001,27(5):651-656. |
| |
| 作者姓名： | 梁彦 谭伟 潘泉 张洪才 |
| |
| 作者单位： | 1.清华大学自动化系,北京; |
| |
| 基金项目： | 国家自然科学基金资助 ( 69772 0 31 ) |
| |
| 摘 要： | 在混合估计中,交互式多模型滤波器(IMM--Interacting Multiple Model)以其优越
的性能而受到广泛的研究.由于马尔可夫参数的限定,交互式多模型在模型数较多时会出现精
度下降,从而限制了它在高维参数空间建模的有效性.利用模型集的概念,首次提出了双马氏
过程的模型切换假设,从而构造出一种两级交互式多模型滤波器.通过辨识系统噪声的多个统
计参数比较了两级交互式多模型滤波器与常规交互式多模型滤波器.结果表明:对于大信噪比
信源(即小的量测噪声),两级交互式多模型滤波器与常规交互式多模型滤波器性能基本相当;
而对于小信噪比信源(即大的量测噪声),两级交互式多模型滤波器明显优于常规交互式多模型
滤波器.
|
| 关 键 词： | 交互式多模型算法 自适应滤波 噪声辨识 |
| 收稿时间： | 1999-9-22 |
| 修稿时间： | 1999年9月22日 |
A Two-Stage Interacting Multiple Model Algorithm |
| |
| LIANG Yan,TAN Wei,PAN Quan,ZHANG Hong-Cai.A Two-Stage Interacting Multiple Model Algorithm[J].Acta Automatica Sinica,2001,27(5):651-656. |
| |
| Authors: | LIANG Yan TAN Wei PAN Quan ZHANG Hong-Cai |
| |
| Affiliation: | 1.Department of Automatic Control,Tsinghua University,Beijing;Department of Automatic Control,Northwestern Polytechnic University,Xi'an |
| |
| Abstract: | In hybrid estimation, interacting multiple model (IMM) estimator is one of the most cost effective schemes. But IMM will degrade if too many models are chosen. This is due to the limit of Markov parameters. In this paper, a dual Markov switching process is proposed. Then a two stage IMM is given. Simulations show that the new algorithm is better than IMM for large measurement noise and almost the same as IMM for small measurement noise when two abruptly changing statistic parameters of process noise are identified. |
| |
| Keywords: | IMM adaptive filtering noise identification |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
| 点击此处可从《自动化学报》浏览原始摘要信息 |
|
点击此处可从《自动化学报》下载全文 |
|
