后验概率与最小均方误差解结合的GNSS部分模糊度解算策略

GNSS Partial Ambiguity Resolution Based on Posterior Probability and Minimum Squares Error Solution

在GNSS弱定位模型中,常常需要采用模糊度部分解算方法获得满意的定位精度。为了进一步提升定位的准确性,本文从3个方面对模糊度部分解算方法进行改进:①决定模糊度被固定元素的依据不再是完全基于模型的Bootstraping成功率,而是同时基于模型和观测数据的后验概率。②当模糊度部分固定时,未固定的模糊度元素不再取条件浮点解,而是取基于已固定元素的最小均方误差解。③当模糊度完全不固定时,模糊度向量不再取浮点解,而是取最小均方误差解。试验中用一组北斗实测数据检验了原GNSS模糊度部分解算方法和新方法。结果表明,新方法部分固定和不固定历元数量少于传统方法,且定位精度显著提高。

GNSS partial ambiguity resolution (PAR) is usually used in weak GNSS positioning model to gain an acceptable position solution.In order to further improve the positioning precise of PAR,this contribution modifies RAR from 3 aspects.① Whether an element of the ambiguity vector is fixed is determined by the posterior probability instead of the Bootstrapping success rate.For the reason that posterior probability is based on data and model simultaneously,whereas Bootstrapping success rate is only determined by GNSS positioning model.②In the ambiguity partially fixed cases,the minimum squares error (MSE) solution of unfixed elements is calculated by the correlation with other fixed ones.However,traditional PAR treats them as conditional float solutions.③ In the ambiguity totally unfixed cases,the MSE solution of the whole ambiguity vector is used to take the place of its float solution.In the experiment,a set of BDS real observed data are used to testify the property of traditional and new PAR.The results show that the partial fixed and unfixed cases for our new PAR method are less than that for the traditional one,and the position precise is sufficiently improved.