基于自适应衰减系数的VB-AKF滤波算法  

A VB-AKF Algorithm Based on Adaptive Attenuation Parameter

作　　者：黄建军[1] 刘杰[1] 

机构地区：[1]深圳大学ATR国防科技重点实验室,广东深圳518060

出　　处：《新型工业化》2013年第11期27-31,共5页The Journal of New Industrialization

基　　金：国家科技支撑计划(2011BAH24B12);教育部博士点基金(20124408110002);武器装备预研项目(XXXX020602);深圳市科技计划项目(JCYJ20130329105816574)

摘　　要：VB-AKF算法利用变分贝叶斯迭代近似方法获得了量测噪声方差较好的估计,但是存在无法完全适应量测噪声方差动态变化的问题。本文提出了一种基于衰减系数自适应的VB-AKF滤波算法。该算法利用量测数据精度等级对衰减系数进行自适应调整,解决了VB-AKF算法事先设定的衰减系数不能完全适应量测噪声方差动态变化的问题。仿真结果表明,该算法能够快速有效的估计出动态变化的量测噪声方差,并且能够有效的实现数据滤波。The variational Bayesian based adaptive kalman filter (VB-AKF) can give good estimation to unknown measurement noise variance,but it cannot fully adapt to the unknown measurement noise variance dynamics.To make it more applicable,an adaptive attenuation parameter is introduced to adapt the dynamics. The proposed algorithm adjusts adaptively the attenuation parameter by the accuracy category of measurement data. Thus the problem that the presetting constant attenuation parameter in VB-AKF cannot fully adapt to the measurement noise variance dynamics is solved. Simulation results show that the proposed algorithm can estimate the dynamic varying measurement noise variance effectively and efficiently, and achieve an effective data filtering.

关 键 词：VB-AKF 精度等级 自适应衰减系数 量测噪声方差估计 

分 类 号：TN391[电子电信—物理电子学]