Adaptive Differential Evolution With Evolution Memory for Multiobjective Optimization

In this paper, a multiobjective differential evolution (MODE) algorithm is developed by incorporating the memory mechanism of particle swarm optimization. That is, the personal best concept is used in the MODE to memorize the evolution of each solution through maintaining a set of non-dominated solutions found by each solution. Besides the adaptive selection of multiple mutation operators that are often adopted in the MODE, an adaptive refining method is used to improve the global external archive. The MODE is referred to as the adaptive MODE with evolution memory (AMODEEM). A set of 30 benchmark problems selected from the literature are used to evaluate its performance. The computational results illustrate that the proposed AMODEEM is competitive or even superior to several powerful MODEs in the literature for most problems.