基于粗差出现的频率信息,建立了观测污染分布模型和粗差数目二项分布模型,讨论了超定方程组基础解粗差分群的性质。通过对基础解解集进行聚类分析,提取基础解解集的零粗差分群,并提出了抗差高斯-雅克比组合平差法。以GPS伪距单点定位为例,利用观测方程基础解的零粗差分群进行了抗差高斯-雅可比组合平差。算例表明,即使杠杆观测为粗差观测,抗差高斯-雅可比组合平差法仍具有高效性和稳健性。
Traditionally,the process of gross error detection is closely associated with parameter estimator.Unreliable initial parameters may cause the process of gross error detection to fail,and vice versa.In this paper,the mixed normal distribution of observation and the binomial distribution of gross error number are introduced by using the prior occurrence frequency of gross error;The paper discusses the properties of gross-error clusters composed of basic solutions,and further propose a robust Gauss-Jacobi combinatorial adjustment method on the zero-gross-error cluster.At last,the proposed method is applied to GPS pseudo-ranging positioning.show that the proposed method is robust and high efficient even gross error occurs on leverage observation.