The experimental design in tasks of interval models identification

Abstract

The technique of optimal experimental design based on I/sub G/- and I/sub E/-criteria for problems of linear multifactor interval model identification is presented. It is shown that I/sub G/-, I/sub E/-optimal saturated designs are orthogonal and equivalent to I/sub D/- and I/sub A/-optimal saturated designs. The consequent design procedure of optimal experiments for synthesis of a linear interval model is considered.