动力学模型系统误差及其协方差阵的随机加权拟合法  

Random weighting fitting method of systemic errors and covariance matrices in dynamic model

作　　者：冯志华[1,2] 高社生[1] 陈丽容[2] 焦雅林[1] 

机构地区：[1]西北工业大学自动化学院,陕西西安710072 [2]北京计算机技术及应用研究所,北京100854

出　　处：《系统工程与电子技术》2012年第2期348-352,共5页Systems Engineering and Electronics

基　　金：国家高技术研究发展计划(863计划)(2010AA7042007)资助课题

摘　　要：在现有的基于移动窗口函数模型和随机模型系统误差自适应拟合方法的基础上,提出一种基于移动窗口动态导航模型系统误差的随机加权拟合法,在相同的窗口内给出了相应的状态预报向量协方差阵的随机加权拟合。由于动力学模型系统误差难以直接修正,采用修正状态估计误差向量及动力学模型误差向量的方法,实现对动力学模型系统误差的修正,然后利用修正后的动力学模型及相应的协方差阵进行导航滤波计算,有效地抑制动力学模型系统系统误差的影响,提高导航解算的精度。仿真结果证明,采用随机加权拟合后的算法精度优于未进行拟合的卡尔曼滤波和自适应卡尔曼滤波算法。On the basis of adaptive fitting method which is based on the existing mobile window function model and the stochastic model system error, a random weighting fitting method for both the systematic errors and covariance matrices of kinematics navigation model errors is presented by using moving windows. The random weighting estimation for covariance matrices of predicted states are given within the same window; the covariance matrices of the modified predicted states are also estimated. The predicted states are then modified. It is shown by the calculation and simulation results that the random weighting estimation algorithm can effectively resist the influence of the systematic errors on the estimated states of navigation, and the performance is superior to the traditional method without random weighting estimation algorithm.

关 键 词：动态模型系统误差 协方差矩阵 随机加权估计 卡尔曼滤波 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]