Adaptive simulation budget allocation in Simulation assisted Differential Evolution algorithm

Abstract

Abstract Recently, the literature on simulation assisted optimization for solving stochastic optimization problems has been considerably growing. In the optimization context, the population based meta-heuristics algorithms, such as, Differential Evolutionary (DE), has shown tremendous success in solving continuous optimization problems. While in the simulation context, Monte-Carlo Simulation for sample average approximation is one of the successful approaches in handling the stochastic parameters of such problems. However, the intertwined computational burden, when combining these two approaches is amplified, and that encourages new research in this topic. In this problem, the challenge is to maintain high quality stochastic solutions by minimizing the computational cost to a reasonable level. To deal with this challenge, we propose a novel Adaptive Segment Based Scheme (ASBS) algorithm, for allocating the MCS budget in a Simulation assisted Differential Evolution (Sim-DE) Algorithm. This allows the algorithm to adaptively control the simulation budget based on a performance measure. The performance of the proposed ASBS algorithm is compared with other simulation budget allocation techniques while using the same base algorithm. The experimental study has been conducted by solving a modified set of IEEE-CEC’2006 test problems and a wind-thermal power systems application. The experimental results reveal that the ASBS algorithm is able to substantially reduce the simulation budget, with an insignificant effect in solution quality.