质点权重对多信息融合姿态估计算法精度的影响研究

The Effect of Particle Weights on the Accuracy of Multi-information Fusion Attitude Estimation Algorithms

首先介绍了一种新的多信息融合的矢量惯性张量姿态估计算法。该算法将每个参考矢量和相关观测矢量的顶端放置一个质点,对质点进行权重分配并计算出此质点系的惯性张量;对每个惯性张量对角化得到特征值、特征向量和主轴坐标系与各坐标系之间的变换矩阵;通过找到各自主轴坐标系的变换,就可以计算出载体坐标系与惯性坐标系之间的方向余弦矩阵,从而得到姿态估计值。质点权重分配对该算法的结果有重要作用,研究了如何给多传感器信息融合的姿态估计算法进行质点权重分配。通过分析,发现当传感器的观测误差接近时,对应的质点应分配相同的权重;而观测误差不同时,精度高的观测矢量应占有较大的分配权重,这样才能得到最优的姿态估计。研究结果对姿态估计问题具有一定的价值。

In this paper,a new multi-information fusion algorithm for vectorial inertial tensor attitude estimation is first introduced.The algorithm places a particle at the top of each reference vector and associated observation vector,assigns weight to the particle and computes the inertial tensor of the particle system;diagonalizes each inertial tensor to obtain the eigenvalue,eigenvector,and transformation matrix between the principal axis coordinate system and each coordinate system;by finding the transformation of the respective principal axis coordinate system,the directional cosine matrix between the vector coordinate system and the inertial coordinate system can be computed.In this paper,we study how to assign prime weights to an attitude estimation algorithm for multi-sensor information fusion.In this paper,we study how to assign prime weights to an attitude estimation algorithm for multi-sensor information fusion.Through the analysis,it is found that when the observation errors of sensors are close to each other,the corresponding prime points should be assigned the same weight;while when the observation errors are different,the observation vector with high accuracy should occupy a larger assigned weight in order to get the best attitude estimation.The results of the study are valuable for the attitude estimation problem.