Memory-based immigrants for ant colony optimization in changing environments

Abstract

Ant colony optimization (ACO) algorithms have proved that they can adapt to dynamic optimization problems (DOPs) when they are enhanced to maintain diversity. DOPs are important due to their similarities to many real-world applications. Several approaches have been integrated with ACO to improve their performance in DOPs, where memory-based approaches and immigrants schemes have shown good results on different variations of the dynamic travelling salesman problem (DTSP). In this paper, we consider a novel variation of DTSP where traffic jams occur in a cyclic pattern. This means that old environments will re-appear in the future. A hybrid method that combines memory and immigrants schemes is proposed into ACO to address this kind of DTSPs. The memory-based approach is useful to directly move the population to promising areas in the new environment by using solutions stored in the memory. The immigrants scheme is useful to maintain the diversity within the population. The experimental results based on different test cases of the DTSP show that the memory based immigrants scheme enhances the performance of ACO in cyclic dynamic environments.