A novel surrogate-assisted differential evolution for mixed-integer expensive constrained optimization problems
Yuanhao Liu, Zan Yang, Danyang Xun, Haobo Qiu, Liang Gao · IET conference proceedings. · 2024
Recently, surrogates have attracted extensive attention due to their ability to combine with evolutionary algorithms to construct high-performance surrogate-assisted evolutionary algorithms (SAEAs) for solving expensive constrained optimization problems (ECOPs). Nevertheless, most of these algorithms only focus on the problems with only continuous variables. In order to solve ECOPs with mixed-integer variables, this paper develop a novel surrogate-assisted differential evolution. Firstly, the information of elite solutions and bad solutions is simultaneously utilized in the bad solutions-assisted elite solutions mutation operation to improve the convergence efficiency. Secondly, a multiple mutation operation and a diverse population generation operation are designed to enable the population to jump out of the local optimal region. Therefore, the proposed algorithm has the ability to balance the performance between convergence and global optimization for mixed-integer ECOPs. Finally, experimental studies on ten benchmarks proves the performance of the developed algorithm.