基于信息补偿的自适应联邦滤波算法

针对采用标准卡尔曼滤波器必须知道系统噪声统计特性的局限性，研究了一类系统噪声未知情况下的自适应联邦滤波方法，指出了自适应滤波方法应用于联邦结构时应当注意的问题，提出了一种基于信息补偿的自适应联邦滤波算法。SINS／BDS／GPS组合导航系统的仿真结果表明，该方法可以有效抑制系统噪声未知情况下的滤波发散现象，提高了滤波的稳定性和估计性能。

Adaptive Federated Kalman Filtering Algorithm with Information Compensation

Aimed at the system process noise sequences inconsistent with their theoretical covariance, the federated filter using the standard Kalman filterng technique yields poor estimation results or diverges. This paper investigates a key problem of applying the conventional adaptive filtering algorithms to the federated filtering structure. By information compensation of the information sharing coefficients, an adaptive federated filtering algorithm is given for the system with unknown process noise covariance. By using the method, the total process noise covariance matrix is derived by the information sharing factors to weigh the local process noise covariance matrixes, and they are respectively computed in each local filter and adaptively adjusted according to the measurement residual sequences of the local filters. Simulation result for the SINS/BDS/GPS integrated navigation systems with unknown process noise shows that the method improves the estimation accuracy and the filtering stability compared with the standard federated Kalman filter.