Optimal computing budget allocation for the vector evaluated genetic algorithm in multi-objective simulation optimization

Gang Kou,Hui Xiao,Minhao Cao,Loo Hay Lee
DOI: https://doi.org/10.1016/j.automatica.2021.109599
IF: 6.4
2021-07-01
Automatica
Abstract:Motivated by the vector evaluation genetic algorithm (VEGA), this research develops simulation budget allocation rules for the VEGA in solving simulation optimization problems. We formulate the selection problem of the VEGA using the optimal computing budget allocation approach, and derive the asymptotically optimal allocation rule and an easily implementable approximated allocation rule. The efficiency of the propose simulation budget allocation rules is demonstrated via comparing with some existing allocation rules. Furthermore, the proposed allocation rule is integrated with the VEGA to solve the multi-objective simulation optimization problems. The numerical experiments on the benchmarking test problems indicate that the proposed allocation rule can improve the search efficiency of the VEGA in stochastic environment by reducing the average distance towards the true Pareto front and improving the purity of the estimated Pareto front.
What problem does this paper attempt to address?