Outlier measurement analysis with the robust estimation

Abstract

Consistent and outlier measurements coexist in the measurement group in the applicable sciences. The adjustment calculus, is made to obtain the nearest solution for real and is detached measurements as consistent or outlier. Various methods are being used in order to determine the outlier measurements in the measurement group. The conventional solution methods and making solutions in accordance with the least squares method, determine the outlier measurements, however, they have some disadvantages like the fact that the corrections are much effected by the errors, they spread the measurement errors to other corrections of measurements, not more than one outlier measurement can be determined in each solution step and that they remove the measurement which is determined to be outlier from the measurement group depending on the structure of the aim function. These disadvantages of the conventional solution method have brought out the search for other methods in order to determine the outlier measurement groups. The robust estimation method presents a solution method compared to the aim function being less affected by the measurement errors. The robust estimation method makes an iterative solution by redetermining the measurement weights according to weight function gained from the aim function being less affected by measurement errors in the solution made in accordance with the least squares method. A few different estimation methods are being used in redetermining the measurement weights. In this study, the methods used for redetermining the measurement weights are explained by expressing the reasons to use the robust estimation methods in the outlier measurements analysis. In order to match the theoretically explained methods, an application has been made by using the real network data and the applicability of these methods has been searched.