Application of genetic algorithm for response surface modeling in optimal statistical design
Yajing Shen, R.M.M. Chen · 2002
In this paper, a new approach for optimal statistical design to achieve 100% yield at minimum cost is introduced. A kind of genetic algorithm together with a boundary sampling strategy is proposed for the selection of the response surface model function to achieve better accuracy of the model representation and improved computation speed due to reduced number of circuit simulations. The complete optimum statistic design also involves the solution procedure of design centering fixed and variable optimum tolerance assignment. A numerical example is presented to show the effectiveness of the new technique.