核模糊C均值聚类算法优选BDS-3三频组合观测值

目前对全球导航卫星系统(global navigation satellite system, GNSS)三频组合观测值优选的研究,主要集中在全球定位系统(global positioning system, GPS)和北斗二号(beidou navigation satellite system, BDS-2)上,对BDS-3的研究相对较少。为克服以往聚类优选算法中存在的仅适用于类球形簇、聚类数目和初始聚类中心的确定主观性强、对离群点敏感、易陷于局部最优等不足,提出一种改进的核模糊C均值聚类算法,引入核函数与抑制离群点的新距离度量,基于多类广义核极化准则优化核参数,用改进爬山法确定聚类数目与初始聚类中心。然后,以模糊C均值聚类算法为对照进行了对比实验,在短、长两种基线下分别解算组合模糊度。通过对优选所得代表性组合的模糊度固定成功率进行对比分析,验证了该算法的可行性与算法改进的有效性。

Optimal triple-frequency combination observations for BDS-3 derived from a modified kernel-based fuzzy C-means clustering algorithm

At present, researches on the optimization of global navigation satellite system (GNSS) triple-frequency combination observations mainly focus on global positioning system (GPS) and Beidou navigation satellite system (BDS)-2, while researches on BDS-3 are relatively few. In order to overcome the shortcomings of previous clustering optimization algorithms, which are only applicable to spherical clusters, highly subjective in the determination of clustering number and initial clustering centers, sensitive to outliers, and easily trapped in local optimal, a modified kernel-based fuzzy C-means (KFCM) clustering algorithm is introduced. This algorithm improves the above deficiencies by using the kernel function and a new kind of distance measure which can suppress the effect of outliers, optimize Gaussian kernel parameters based on generalized kernel polarization, initiale the number of clusters and the selection of initial clustering centers based on a new kind of modified mountain method. Then the fuzzy C-means clustering algorithm is used as a comparison experiment. Under short and long baseline conditions, the combination ambiguity is calculated respectively. By comparing and analyzing the ambiguity fixed success rate of the selected representative combinations, the feasibility of the algorithm and the effectiveness of the improved algorithm are proved.