势函数解决附不等式约束平差问题的研究
首发时间:2006-08-16
摘要:测量数据处理中经常会有些先验信息可以利用,这些先验信息可以总结成等式或不等式。附等式约束的平差理论目前已经十分成熟,因而如果是等式约束,则可用附等式约束的间接平差方法来处理。但如果是不等式约束,则计算相对困难。Frisch(1955)和Carroll(1961)等人相继提出的将不等式约束转化为无约束的罚函数算法在求解优化问题上取得了非常好的效果,然而这种经典的罚函数随着惩罚因子的增大,其Hessian矩阵会出现病态,收敛速度会变的很慢。本文试图运用罚函数的思路,尝试引入一种新的势函数算法并结合最小二乘平差模型来解决附不等式约束的平差问题;文中通过分析基于最优性条件(K-T条件)下该势函数的性质来推导平差结果及其统计特征的显式表达式,从数值算例的结果证明该算法具有可行性。
For information in English, please click here
Solving the LICA Problem by a Potential Function Method
Abstract:In survey data processing, sometimes there is prior information that can be used. This prior information can be expressed by equality or inequality constraints on the parameters. The equality-constrained adjustment theory has been maturated and comprehended, so if the constraints are equality the problem can be solved by the equality-constrained indirect adjustment theory. However, if the constraints are linear inequality, the problem become linear inequality constrained adjustment (LICA) and the computation will be very difficult. Frisch(1955) and Carroll(1961) proposed that penalty function which transformed constrained-unequation into unstrained function obtain good result, but the Hessian matrix will be sick and the convergence speed will be slow with the penalty factorial\
Keywords: potential function LICA least-square adjustment explicit expression
论文图表:
引用
No.7938104891155718****
同行评议
共计0人参与
勘误表
势函数解决附不等式约束平差问题的研究
评论
全部评论0/1000