Enhancement of enhanced stochastic evolutionary algorithm for computer simulated experiment

Abstract

Computer simulated experiments (CSE) are currently used to replace complex engineering systems and applications. The design of experiment plays the major roles in CSE, since the accurate surrogate model usually obtains from the best experimental design. The process of constructing the design in this context is a space fi lling proces, so search techniques along with different optimal criterion are applied. This paper purposes the enhancement of enhanced stochastic evolutionary (EESE), adapted from the enhanced stochastic evolutionary (ESE), to search or construct the optimal latin hypercube design (LHD) based on Φp criteria. The results show EESE performs better than ESE in any dimensional designs, especcially in a small dimension in terms of time and number of elementexchange.