研究适用于非线性不确定系统的多模型自适应估计(MMAE)算法, 基于一系列不同的系统噪声方差阵建立多个模型, 使得由多个并行滤波器构成的算法适用于存在未知干扰的导航系统. 针对非线性不确定系统, 分析了多模型自适应估计算法的稳定性, 给出了算法误差有界的充分条件. 重点研究用于远距离航天器相对导航的MMAE算法, 精确获取空间目标的相对位置, 对于执行空间任务或实现航天器避碰具有重要意义. 为了改善导航系统的可观度, 采取双视线测量目标跟踪策略. 以基于双视线矢量测量的空间目标相对导航问题为例, 对扩展卡尔曼滤波(EKF), 非线性鲁棒卡尔曼滤波(NRF)和MMAE算法进行了对比研究, 结果表明, 系统中存在未知干扰的情况下, 所设计的MMAE算法具有优于传统EKF和NRF的性能.

This paper studies the multiple model adaptive estimator (MMAE) for nonlinear systems with unknown disturbances. Multiple models are constructed with a set of process noise covariance matrices, such that the algorithm that consists of multiple parallel filters can adapt to different levels of unknown disturbances. The filtering stability of the MMAE is analyzed for the considered system. Sufficient conditions to ensure the boundedness of the algorithm is provided. Then we focus on the utilization of the MMAE for spacecraft relative navigation missions. The ability to know precise position of a space target is critical to mission planners for reaching science objectives as well as avoiding catastrophic collisions with other spacecrafts or space debris. In order to improve the observability of the system, the double LOS measuring target tracking technique is used. A performance comparison among an extended Kalman filter (EKF), a nonlinear robust filter (NRF) and the MMAE is carried out for spacecraft relative navigation, where the position of a space target is estimated by using double line-of-sight (LOS) measurements. Simulation studies illustrate that the MMAE performs better than the EKF and the NRF.