Implementation of nonlinear least square with global convergence in Forstat

The Gauss-Newton method is applied to solve the nonlinear least square equations and a simple and applicable iterative formula is deduced, which is locally convergent and divergent sometimes.

Through transferring solution procedure,the problem was inverted into nonlinear least square optimization process with constraint conditions in two stages using function lambertw and function lsqnonlin.

Algorithm of nonlinear least squares adjustment of any plane networks with coordinates computed automatically;

A new newton iterative algorithm for solving nonlinear least squares problem;

Unified model and ill-posed property of numerical iterative formula for solving nonlinear least squares problem;

For fixed targets and moving constant velocity targets, deterministic parameters calculation and linear least square algorithms are applied to set up mathematic model and to do some algorithm simulations.

This paper analyses the systemetic error of calculating formula of extremum detection depth method of undergound metal pipe detector,fites for experimental data by means of linear least square method and obtains the modifying factor of the formula.

On the basis of the homotopy arithmetic, this paper puts forward a uniform model of nonlinear least square (LS) adjustment, which can be used not only for the nonlinear LS adjustment of the rank defect problems, but also for that of the rank full problems.