A simplex search method for experimental optimization with multiple objectives

MARTINEZ, ERNESTO

Computer-aided Chemical Engineering

Elsevier

Lugar: Amsterdam; Año: 2006; p. 377 - 382

Experimental optimization with multiple responses in the face of noise and outliers is akey issue in product/process development and to guarantee product end-use properties.A novel concept of maximimizing the concordance of desirability functions of allresponses concerned is presented. The Kendall´s coefficient of concordance Wborrowed from nonparametric statistics is used to provide a statistical characterizationof optimality in a multiobjective setting. A multi-directional simplex method isproposed in the concordance function framework. Simplex reflection, expansion andcontraction operations are based on ranking vertices according to their correspondingvalues of W.