衰减因子自适应估计卡尔曼滤波比较研究  

Comparative Study on Adaptive Fading Kalman Filter

作　　者：耿延睿[1] 李大字[1] 郭文荣[2] 

机构地区：[1]北京化工大学自动化研究所,北京100029 [2]北京联合大学自动化学院,北京100101

出　　处：《控制工程》2006年第S2期70-72,共3页Control Engineering of China

基　　金：教育部留学回国人员科研启动基金;北京市教育委员会共建项目建设计划基金(XK100100435)

摘　　要：针对卡尔曼滤波算法发散的问题,从卡尔曼滤波技术的稳定性出发,分析了卡尔曼滤波发散的原因,提出了新的衰减记忆卡尔曼滤波中衰减因子的自适应估计方法。该方法利用滤波残差序列在最优估计时为零均值白噪声的性质,分别检验滤波残差每一个分量得出衰减因子值,并与强跟踪滤波器进行了对比研究。仿真结果表明,新算法在系统噪声特性不准确的情况下,能自适应地估计出衰减因子的大小,抑制卡尔曼滤波估计的发散,滤波精度要高于强跟踪滤波器;且其推导形式简单、计算量小、适合于在线运算。The reasons of the instability of Kalman filter are analyzed from the stability of Kalman filter.A new approach to adaptive estimation of Kalman filter fading factor is developed and is compared with the strong tracking Kalman filter.The characteristic that the filter residuals are zero 2 mean Gaussian white noise vectors is used and a chi 2 square distribution variable is made while computing the fading factor.Simulation result shows that the proposed method has the ability of restraining filtering divergence under the condition of wrong system noise attributes and has better efftect of estimation.The derivation of new method is simple and the computation burden is low enough to adapt to calculate online.

关 键 词：卡尔曼滤波 自适应滤波 强跟踪滤波 衰减记忆滤波 

分 类 号：TN713[电子电信—电路与系统]