Value of information method for optimization and experimental design using surrogate models

Experimentation is required for modeling empirical functions and optimization. In manufacturing, experiments are costly and time-consuming, thereby limiting the number of function evaluations. This paper describes a value of information method for experimental design and optimization using surrogate modeling. Value of information is defined as the absolute difference between optimal value before experiment and the expected optimal value after experiment, or, the expected improvement in the optimum after experiment. The value of information based experimental design performs better than the traditional statistical design of experiments such as Taguchi orthogonal arrays, and central composite design, especially in three or more dimensions.