Métodos de superficie Multirespuesta: Un Estudio Comparativo

Abstract

The simultaneous optimization problem may not to have a complete satisfactory solution from the point of view of the individual responses, in the sense that individual optimums are different respect to global optimum; but always it is possible to say that it exists the process operation conditions (point in the factors space) where the responses fit “in the best way” to their specification limits and target values. It is always possible to obtain a compromise solution, which look for the best balancebetween the responses. This paper discusses several methods that have been proposed for analyzing multi-response data, and it is shown that the graphical method can raise the best solution compared with the analytical methods. The performance of the methods is compared in the context of one example. Finally, in two of the methods we suggest alternative weighting of the responses in order to improve the results.