Immune Generalized Differential Evolution for dynamic multi-objective environments: An empirical study

In this paper, an Immune Generalized Differential Evolution 3 (Immune GDE3) algorithm to solve dynamic multi-objective optimization problems (DMOPs) is empirically analyzed. Three main issues of the algorithm are explored: (1) the general performance of Immune GDE3 in comparison with other well-known algorithms, (2) its sensitivity to different change severities and frequencies, and (3) the role of its change reaction mechanism based on an immune response. For such purpose, four performance metrics, three unary and one binary, are computed in a comparison against other state-of-the-art dynamic multi-objective evolutionary algorithms (DMOEAs) when solving a novel suite of test problems. A proposal for the adaptation of a binary metric, called Two-set-coverage, to evaluate the performance of DMOEAs is also presented in this paper. The statistically validated results indicate that Immune GDE3 is robust to change frequency and severity variations and can track the environmental change finding a good distribution of solutions. Finally, Immune GDE3 has a very competitive performance solving different types of DMOPs and this good performance is mainly attributed to its change reaction mechanism based on an immune response. Numerical results support such findings, showing that Immune GDE3 obtains good results in all performance metrics, especially in the distribution metrics: Spacing(S) and Two-set-coverage(C-metric).