基于容错型联邦强跟踪滤波的多星座组合导航算法研究

摘要

When the multi-constellation integrated navigation system is applied to the high-dynamic environment,with the influence of the acceleration changing in a wide range and the properties for kinematics noise and measurement noise difficult to be ascertained exactly,the estimation precision of the general federated filtering decreases seriously,and the filtering even became divergence.Regarding to this problem,this paper applies strong tracking filter to fault-tolerant federated filter,forming a fault-tolerant federated strong tracking filter（FTFSTF）.The algorithm proposed is applied on COMPASS/GPS/GLONASS multi-constellation integrated navigation system.The simulation results show that the algorithm is flexible,strong fault tolerant,and better to track high-dynamic targets,and has significantly improved the precision and reliability of the multi-constellation integrated navigation system.Simultaneously,owing to the combined application of the no-reset mode federated filter and some practical algorithms including the one-step suboptimal calculation method for fading matrix and the residual Chi-square detecting method,the FTFSTF algorithm has moderate calculation complexity,and it is easily implemented and valuable for engineering application.%在多星座组合导航应用于高动态场景时,由于受加速度变化范围大、动态噪声和观测噪声难以准确预测等因素影响,常规联邦滤波估计精度将会严重下降甚至发散。针对这一问题,提出将强跟踪滤波算法应用到容错型联邦滤波器中,构成容错型联邦强跟踪滤波器。对COMPASS/GPS/GLONASS组合导航系统进行的仿真结果表明：该算法设计灵活,容错性强,对高动态目标有较强的跟踪能力,能够显著提高导航定位的精度和可靠性。同时,由于组合应用了无重置联邦滤波结构和渐消矩阵一步次优算法、残差χ2检验算法等实用算法,使得该算法整体计算量适中,易于实现,具有一定的工程实用价值。