A Hybrid Learning-Based Differential Evolution for Mixed-Variable Optimization Problems
Huimin Che, Wei Xing Zheng, Mingming Yang · 2024
Mixed-variable optimization problems (MVOPs) involve both discrete and continuous decision variables which lead to a lot of difficulties solving these problems. In this paper, a hybrid differential evolution for MVOPs, termed HDEmv, is proposed. In HDEmv, K-means clustering is used for learning population structure so that promising mating parents for each individual can be selected. Moreover, an operator pool with adaptive parameters is constructed to produce offspring and a simple local search method is employed to fine tune the discrete variables. In experiments, HDEmv is compared with four promising algorithms on 28 MVOPs and the final results reflect that HDEmv is a competitive method for MVOPs.