快速交互式多模型算法的导航定位解算

A fast IMM algorithm in the application of the navigation and positioning

针对卫星导航定位中单模型在急速转弯条件下出现较大偏差和交互式多模型系统运算量大的问题,该文提出了一种快速的交互式多模型算法。该算法先对CS模型进行改进,然后利用CV模型和改进的CS模型构成多模型系统,结合修正的滤波发散判据与次优的极大后验估计,根据载体实际的运动状态自行调整扩展卡尔曼滤波采用的运动模型,从而弥补了单模型描述复杂运动的不足。实验结果表明:与单模型相比,该方法有效解决了急速转弯误差大的问题;与标准交互式多模型相比,该方法对位置的估计精度提高了39.4%,而运算时间却缩短了47.4%。

Aiming at the problem which single model appeared large deviation under the condition of rapid turn and the large computation of interactive multiple model system in satellite navigation and positioning, a fast IMM algorithm was proposed. First, the current statistics（CS）model was improved; then the constant velocity（CV） model and the improved CS model were used to form a multiple model system. Based on the modified filter divergence criterion and the subprime maximum a posteriori estimation, the motion model of extended kalman filter was changed automatically according to the actual movement of carrier. In this way, the shortcoming of single model in describing the complex movement was improved. The experimental results showed that compared with the single model, the method effectively solved the problem of the great error in the sharp turn; compared with standard IMM, the location estimation precision of the method was improved by 39.4%, and the operation time was shortened by 47.4%.