Adjustment Algorithm About Observations With Bounded Constraint

ABSTRACT:In the measurement data acquisition often exist uncertainty, these uncertainties differs from the ordinary error, there is usually no fixed value, and also being unable to determine the statistical information. Traditional adjustment methods based on the theory " Uncertainty is random’ not only failed to better estimate the uncertainty, may even increase the uncertainty of estimated results. Least-squares method is widely used in the field of measurement data processing. It is easy to use, but usually only consider the error of the observation vector, and ignore the error of the coefficient matrix. In recent years, more and more attention to total least squares method, to some degree, it can to reduce the influence of uncertainty, but it have to consider uncertainty of both measurement matrix and coefficient matrix, the overall adjustment criterion may cause excessive correction, especially in the condition that uncertainty of the coefficient matrix and measurement vector have a priori information. So use the uncertainty theory, establish new rule of adjustment method, put forward a better measurement data processing method to moderate the influence of the uncertainty is very necessary.Based on the previous question, the author mainly conducted the following research work:1.Coefficients that may exist for surveying data processing errors in the same column of the matrix with the same constraints, to meet adjustment guidelines for dealing with such peaceful method.2.Studied Neural Network algorithm of observational data with bounded constraints (box-type constraints), and analyze the optimality of the solution.3.Set up the adjustment model with bounded error constraints and rules of general adjustment, that’s the coefficient matrix and the observation vector norm is smaller than a constant function model, studied the corresponding adjustment methods, and analyzed the effect of algorithm in this paper and least-squares and total least-squares, because adjustment in this paper criterion of error bounded to the constraint of a priori information, so the new adjustment method can get the parameters of better results.