On a paired-t confidence interval based ranking and selection method
Simulation optimization is a widely used methodology for complex stochastic system design. Though the simulation provides a flexible tool for model building, the optimization process could be very challenging due to the cumbersome computing times. A paired-t confidence interval based ranking and selection method is developed to tackle such a problem. The survive probability of each candidate system is calculated with the lower and upper bounds of the confidence interval obtained using paired-t comparison method. The proposed method is tested with several numerical examples. The experimental results suggest that the proposed method outperforms the other classical Ranking and Selection procedures.
Includes bibliographical references