卡尔曼体系下的滤波算法计算框架  

Implementation Framework of Filters in Kalman Structure

作　　者：王建文[1] 税海涛[1] 马宏绪[1] 

机构地区：[1]国防科技大学机电工程与自动化学院,长沙410073

出　　处：《数据采集与处理》2009年第1期61-66,共6页Journal of Data Acquisition and Processing

摘　　要：卡尔曼体系下的滤波算法是指滤波算法中含有基于状态方程的状态预测过程和基于观测方程的状态更新过程。为了便于理解卡尔曼体系下的滤波算法计算过程,从滤波算法计算框架角度对它们分别进行了描述。提出了一个统一的卡尔曼体系下的滤波算法计算框架,该统一计算框架既可用于理解滤波算法计算过程又可用于构造新滤波算法。在统一计算框架中存在两个反馈回路,构造新滤波算法的难点在于确定两个反馈回路中的操作函数。本文以自适应卡尔曼滤波算法(Adaptive Kalman filters,AKF)为例就操作函数选择问题进行了初步探讨,证明了几种操作函数是次优的,这为最终构造一种性能优良的AKF算法奠定了良好的理论基础。Filters in the Kalman structure are defined. The filters include a predictor based on the process equation and a corrector based on the measurement equation. To understand imple- mentation processes of filters in the Kalman structure, some filters are described from imple- mentation frameworks. A unified implementation framework is adapted to describe implemen- tation processes of all filters in the Kalman structure. This unified implementation framework can be used to understand implementation processes of filters in the Kalman structure and de- sign novel filters. Two feedback loops exist in the unified implementation framework. Difficulties in designing novel filters are choices of operation functions in two feedback loops. Adaptive Kalman filters （AKFs） are taken as examples, to study choices of operation functions. And a few operation functions are proved to be suboptimal. These conclusions are reliable for designing perfect AKF.

关 键 词：滤波算法 统一计算框架 自适应卡尔曼滤波算法 次优操作函数 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]