基于模糊理论的自适应联邦滤波算法研究

针对先验噪声统计特性与实际不符引起卡尔曼滤波精度下降的情况,提出了一种基于模糊自适应联邦卡尔曼滤波方法。算法基于模糊推理思想通过理论残差与实际残差的比值来对观测噪声协方差阵进行加权调整,实现对观测噪声统计特性的实时跟踪,在此基础上,利用信息分配因子对各子滤波器及主滤波器输出进行加权,得到滤波器的全局输出,将此输出周期性反馈给各个子滤波器以校正子滤波系统的偏差。SINS/GPS/TAN组合导航系统的仿真试验表明该算法具有更强的自适应性、滤波精度更高、鲁棒性更强。

Study of an adaptive federated Kalman filter algorithm based on fuzzy

Considering the divergence phenomenon that the prior noise statistical property of Kalman filter does not agree with its real behavior,an adaptive federated Kalman filter algorithm based on fuzzy is proposed.First,observation noise covariance matrix is weighted with the ratio of theoretical residual and realistic residual based on fuzzy inference system to track statistic of measurement noise on line,on that basis,the results of sub-filter and federated filter are weighted with fator of information distribu...