Immune clonal algorithm based on directed evolution for multi-objective capacitated arc routing problem

In the latter two lines of figures, DE-ICA gets the best convergence on 5 instances (e4a, s1a, s4a, G1-A, G2-E). On these instances, DE-ICA can reach areas with both low total cost and the low makespan. As a result, the non-dominant solutions obtained by DE-ICA almost completely dominate the non-dominant solutions obtained by D-MAENS and ID-MAENS. On the remaining instances, DE-ICA is also not worse than the other two algorithms in convergence. In conclusion, DE-ICA demonstrates an obvious advantage in convergence on these large scale instances. The results above show that the DE-ICA is more suitable for the large-scale problems.