基于动态不确定度评定的加权数据融合算法

A weighted data fusion algorithm based on dynamic uncertainty evaluation

加权融合算法的关键是权值因子的确定。基于"滑动窗"处理方法对定位序列建立AR模型,采用遗忘因子递推最小二乘法估计出测量方差。根据测量方差对单一导航传感器信息进行动态不确定度评定,进而确定组合导航系统的组合不确定度。在满足组合不确定度最小的情况下,利用拉格朗日乘数法进行加权值的求解。通过实例仿真表明该法具有较高的融合精度,对提高组合导航系统的定位精度具有实际意义。

Weight value＇s calculation is of great importance to weighted data fusion algorithm. AR models are built up for positioning sequence based on ＂sliding window＂ method. Measuring variance is evaluated through RLS method based on forgetting factor. Then the dynamic uncertainty of single navigation sensor is evaluated according to this variance, and thus the combined uncertainty of the combined navigation system is worked out, Under the condition of minimum combined uncertainty, the weight values are calculated by Lagrange method. The simulation indicates the algorithm has a high fusion accuracy and realistic significance to improving positioning accuracy of a combined navigation system.