Strategy Adaption Multi-objective Differential Evolution Algorithm with Adaptive Grids-based Diversity Maintenance Method

In order to provide a multi-objective optimization algorithm with high usability and adaptability, and avoid the time-consuming trial and error manual choice of mutation strategies and control parameters, a strategy adaption multi-objective differential evolution algorithm (SAMODE) is formulated in this article. The different perturbation features of six typical mutation strategies are analyzed firstly. Then, through adjusting the mutation vector generation strategy in conjunction with the control parameters at different evolution phases for different individuals, the mutation vector generation strategy adaption mechanism and control parameter adaption mechanism are analyzed successively. Finally, the overall algorithmic framework of the SAMODE is presented, which includes the mutation vector adaption mechanism, parameter adaption mechanism, external archive update rules and adaptive grids mechanism. The experimental results show that incorporating the strategies adaption mechanism into the MODE improves the performance of the results effectively on most test functions.