半参数模型在大高差短基线解算中的应用研究

发布时间：2018-04-28 23:34

本文选题：GPS + 半参数　；参考：《合肥工业大学》2016年硕士论文

【摘要】：高精度的GPS定位一般采用双差技术来消除或削弱各种误差,提高基线解算精度。当两站高差较大时,即使是短基线,通过双差技术也不能充分消除对流层延迟,残余对流层误差对基线解算的影响较大。由高差带来的残余对流层延迟较大时,模糊度浮点解误差大,影响模糊度固定,甚至可导致基线解算失败。在参数模型中,无法采用参数估计其影响,如果采用现有的参数模型来描述观测值,不可避免的存在模型误差。本文提出采用半参数回归模型解算大高差短基线,削弱残余对流层误差影响。实际算例表明,与常规的最小二乘法相比,采用半参数模型处理基线能很好的分离GPS残余对流层延迟误差,提高基线解算精度。本文主要研究内容包含以下几个方面：1.GPS静态相对定位半参数模型。在模糊度固定的基础上,采用半参数模型对大高差短基线进行解算。运用半参数模型关键是选取合适的正则矩阵R和正则化参数α。本文正则化矩阵R采用时间序列法确定,正则化参数α采用L曲线法确定。对流层延迟改正模型选用Saastamoinen模型,映射函数选用NMF函数。实际算例表明,与常规的最小二乘法相比,采用半参数模型处理基线能很好的分离GPS残余对流层延迟误差,提高基线解算精度。2.GPS单历元模糊度解算。针对单历元解算大高差短基线时载波相位模糊度难以固定或者固定错误的情况,采用部分搜索算法求解模糊度。实际算例表明与常规LAMBDA方法相比,该算法具有一定的优越性。3.GPS单历元相对定位半参数模型。采用半参数广义补偿最小二乘模型对大高差短基线进行单历元解算,运用部分搜索法并结合LAMBDA方法求解载波相位模糊度,岭参数采用基于Helmert方差分量估计的岭参数法确定。实例验证表明,与常规的最小二乘法相比,采用半参数广义补偿最小二乘单历元基线解算能很好的分离GPS残余对流层误差,各方向定位精度达到毫米级水平。
[Abstract]:High precision GPS positioning usually uses double difference technique to eliminate or weaken all kinds of errors and to improve the accuracy of baseline solution. When the height difference between the two stations is large, even if the short baseline is short, the tropospheric delay can not be fully eliminated by the double difference technique, and the residual tropospheric error has a great influence on the baseline solution. When the residual tropospheric delay caused by height difference is large, the error of ambiguity floating point solution is large, which affects the ambiguity fixed, and even leads to the failure of baseline solution. In the parameter model, it is impossible to use the parameter to estimate its influence. If the existing parameter model is used to describe the observed value, the model error will inevitably exist. In this paper, a semi-parametric regression model is proposed to solve large height difference and short baselines, which weakens the influence of residual tropospheric errors. A practical example shows that compared with the conventional least square method, the semi-parametric model can effectively separate the residual tropospheric delay error of GPS and improve the accuracy of the baseline solution. The main contents of this paper include the following aspects: 1. GPS static relative positioning semi-parametric model. On the basis of fixed ambiguity, the semi-parametric model is used to calculate the short baseline with large height difference. The key of using semi-parametric model is to select appropriate regular matrix R and regularization parameter 伪. In this paper, the regularization matrix R is determined by time series method, and the regularization parameter 伪 is determined by L curve method. Saastamoinen model is used for tropospheric delay correction model and NMF function is used for mapping function. The practical example shows that compared with the conventional least square method, the half-parameter model can deal with the residual tropospheric delay error of GPS and improve the precision of baseline resolution. 2. The ambiguity resolution of single epoch of GPS is improved. In view of the fact that the ambiguity of carrier phase is difficult to be fixed or fixed when the single epoch is used to calculate the large height difference and short baseline, the partial search algorithm is used to solve the ambiguity. A practical example shows that the algorithm has some advantages over the conventional LAMBDA method. 3. The semi-parametric model of relative positioning of single epoch is obtained. The semi-parametric generalized compensated least square model is used to calculate the single epoch of large height difference and short baseline, and the partial search method combined with LAMBDA method is used to solve the ambiguity of carrier phase. The ridge parameter is determined by the ridge parameter method based on Helmert variance component estimation. The example shows that compared with the conventional least square method, the GPS residual tropospheric error can be separated well by using the semi-parameter generalized compensated least square single epoch baseline solution, and the positioning accuracy in each direction reaches millimeter level.
【学位授予单位】：合肥工业大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：P228.4

【相似文献】