基于自适应联邦滤波的卫星姿态确定

卡尔曼滤波采用常值量测噪声协方差阵，当量测噪声统计特性发生变化时，易导致估计误差增大，甚至滤波发散。针对该问题，在联邦卡尔曼滤波子系统中采用自适应卡尔曼滤波，形成自适应联邦卡尔曼滤波算法，新算法采用模糊推理系统实时调整量测噪声协方差阵的加权系数,使模型量测噪声逐渐逼近真实噪声水平。将该算法应用于多传感器卫星姿态确定系统，仿真结果验证了算法的有效性。

Satellite Attitude Determination Based on the Adaptive Federated Kalman Filter

Standard Kalman filter adopts constant covariance of measurement noise. When statistical characteristics of measurement noise changes, estimation error increases, which results in filtering divergence. An adaptive federated Kalman filter was proposed with fuzzy adaptive Kalman filter but not Kalman filter in the subsystem of federated Kalman filter, and the weighted coefficient of covariance matrix was adjusted by fuzzy inference algorithm real-timely. It made the measurement noise of the dynamic equation close to the truth level. When it is applied to multi-sensor attitude determination systems, simulation results demonstrate the true effectiveness of the proposed algorithm.