Generalized Test Suite for Continuous Dynamic Multi-objective Optimization

Chang Shao, Qi Zhao, Yuhui Shi, Jing Jiang

Abstract: Dynamic multi-objective optimization (DMO) has recently attracted increasing interest. Suitable benchmark problems are crucial for evaluating the performance of DMO solvers. However, most of the existing DMO benchmarks mainly focus on Pareto-optimal solutions (PS) varying on the hyperplane, which may produce some unexpected bias for algorithmic analysis. Furthermore, they do not comprehensively consider the general time-linkage property, yet which is commonly observed in real-world applications. To alleviate these two issues, we designed a generalized test suite (GTS) for DMO with the following two advantages over previous existing benchmarks: 1) the PS can change on the hypersurface over time, to better compare the tracking ability of different DMO solvers; 2) the general time-linkage feature is included to systemically investigate the algorithmic robustness in the dynamic environment. Experimental results on five representative DMO algorithms demonstrated the proposed GTS can efficiently discriminate the performance of DMO algorithms and is more general than existing benchmarks.