平坦地形条件下改进RPF的TAN方法

TAN method for improving RPF under flat terrain conditions

针对地形辅助导航系统中递归地形匹配方法在平坦地形条件下位置估计鲁棒性差的问题,提出了一种基于集合卡尔曼滤波和正则化粒子滤波(RPF)的地形匹配方法。首先分别以航行器的水平位置分量和多波束声纳的高程测量值作为地形匹配系统的状态量和观测量,然后采用基于投影的方案补偿航行器姿态变化导致的测深误差,最后利用集合卡尔曼滤波器更新RPF中的条件建议分布以实现递归地形匹配。通过船载湖试数据评估了改进RPF在不同初始匹配位置误差条件下的地形匹配跟踪性能,结果表明:所提地形匹配滤波器能始终保持有界的定位误差,位置跟踪精度和置信区间估计性能较高,在10 m分辨率的先验数字地形图中地形匹配误差均值小于2个网格。

To address the problem of poor position estimation robustness of the recursive terrain matching methods in the terrain aided navigation system under flat terrain conditions,a terrain matching method based on ensemble Kalman filter and regularized particle filter(RPF)is proposed.Firstly,the horizontal position component of the vehicle and the elevation measurement value of multi-beam sonar are used as the state and measurement variables of the terrain matching system,respectively.Then,the projection-based scheme is adopted to compensate for depth errors caused by attitude changes of the vehicle.Finally,the ensemble Kalman filter is used to update the conditional proposal distribution in RPF for recursive terrain matching.The terrain matching tracking performance of the improved RPF is evaluated by using ship-borne lake test data under different initial matching position error conditions.The results show that the proposed terrain matching filter can always maintain bounded positioning errors,and has high position tracking accuracy and confidence interval estimation performance.The average terrain matching error is less than 2 grids in a prior digital terrain map with a resolution of 10 m.