Root
/
Conference Proceedings
/
4th International Conference on Physics and Control (PhysCon 2009)
/
A New Linear Programming Algorithm for Optimal Estimation and Correction of an Aircraft Trajectory
/

Problems of optimal estimation under uncertainty and scalar optimal design of experiment may be reduced

to a problem of linear programming. Problems of ideal linear ideal trajectory correction and many problems of optimal design of experiment may be solved with the help of multiparametrical (so-called generalized) linear programming. Traditional methods for solving these problems are the simplex method and the the column-generate method

respectively. While using these methods two main difficulties may appear. Current basic matrix often turns out to be be ill

conditioned, and the solution is usually accompanied by a large number of almost degenerate iterations thus accumulating large computational errors. Besides this convergence of the column-generate method is not proved. An new efficient method for solving problems of this class the so-called skeleton algorithm is proposed. This algorithm helps to avoid problems mentioned above. The algorithm is easy

enough because it does not use inversion of matrices.

to a problem of linear programming. Problems of ideal linear ideal trajectory correction and many problems of optimal design of experiment may be solved with the help of multiparametrical (so-called generalized) linear programming. Traditional methods for solving these problems are the simplex method and the the column-generate method

respectively. While using these methods two main difficulties may appear. Current basic matrix often turns out to be be ill

conditioned, and the solution is usually accompanied by a large number of almost degenerate iterations thus accumulating large computational errors. Besides this convergence of the column-generate method is not proved. An new efficient method for solving problems of this class the so-called skeleton algorithm is proposed. This algorithm helps to avoid problems mentioned above. The algorithm is easy

enough because it does not use inversion of matrices.