AJUSTE POSINOMIAL Y SIGNOMIAL MEDIANTE LA GENERACIÓN DE MONOMIALES POR MÉTODOS DE PROGRAMACIÓN NO LINEAL

Abstract

The posynomial and signomial adjustments of functions have been used in engineering and other scientific areas because they allow to model and to solve problems of non-linear and non-convex optimization by applying of the geometric programming model or similar to it. The present work had as objective to propose new techniques of generation of locally optimal monomials using the combination of the methods of the minimum square adjustment and of non-lineal programming to improve the quality of monomials generated in search process of parameters. The quality of monomials is measuring by the distance between the best monomial obtained and the original function. For the comparison of the obtained results, two numeric examples of monomial, posynomial and signomial functions adjustment with one and two independent variables were extracted from previously developed works and have showed that the new proposed techniques improve the quality of locally optimal monomials.