测绘通报 ›› 2018, Vol. 0 ›› Issue (8): 37-40,46.doi: 10.13474/j.cnki.11-2246.2018.0241

• 学术研究 • 上一篇    下一篇

测距定位方程的多解性及其非线性最小二乘迭代算法

齐珂1,2, 曲国庆1, 薛树强2, 刘以旭1, 杨文龙1, 韩德强2   

  1. 1. 山东理工大学, 山东 淄博 255049;
    2. 中国测绘科学研究院, 北京 100830
  • 收稿日期:2017-11-27 修回日期:2018-01-09 出版日期:2018-08-25 发布日期:2018-08-30
  • 作者简介:齐珂(1992-),男,硕士,主要研究方向为水下定位控制网优化与非线性定位算法。E-mail:805476058@qq.com
  • 基金资助:
    国家自然科学青年基金(41704003);国家自然科学基金面上项目(41674014)

Multiple Solutions of Distance Equations and Nonlinear Least Squares Iterative Algorithms

QI Ke1,2, QU Guoqing1, XUE Shuqiang2, LIU Yixu1, YANG Wenlong1, HAN Deqiang2   

  1. 1. Shandong University of Technology, Zibo 255049, China;
    2. Chinese Academy of Surveying and Mapping, Beijing 100830, China
  • Received:2017-11-27 Revised:2018-01-09 Online:2018-08-25 Published:2018-08-30

摘要: 距离观测在测量中具有重要的地位,其观测方程为非线性函数模型。由于非线性问题的复杂性,测距方程的解可能不是唯一的,尤其当方程呈现病态性时,方程的解将会变得更加复杂,同时短距离测距方程的非线性强度相对较大,会对迭代算法产生影响。本文针对这一问题,利用直接解法、高斯-牛顿法和封闭牛顿法对病态的测距方程进行求解,试验验证了解析法、高斯-牛顿法与封闭牛顿迭代法的局部收敛性质。结果表明封闭牛顿法的局部收敛性最好。探讨了非线性定位问题中的病态多解问题,解析法和高斯-牛顿法会得到两个解,封闭牛顿法会得到3个解,并且其中两个解与前两种算法相同,而且这些解均为局部最优解。

关键词: 测距方程, 非线性, 多解性, 迭代算法

Abstract: Distance observations play a very key role in surveying,of which the related observation model is nonlinear.Because of the complexity of nonlinear problems,observation equations may have a non-unique solution,particularly when the equation is ill-conditioned,the solutions of the equation may become more complex.In view of this problem,the paper uses the direct method and Gauss-Newton method and closed-form of Newton method to solve ill-conditioning,testing the their performances and local convergence of three methods.The result shows that the closed-form of Newton method has the best local convergence.In the experiment,the direct method and Gauss-Newton method can get two solutions,while the closed-form of Newton method can get three solutions two of which are the same with the solutions of direct method and Gauss-Newton method,but,the search total solutions are the best local optimum.

Key words: distance observation, non-linear models, multiple solutions, iterative method

中图分类号: