Ant Colony Optimization with Immigrants Schemes in Dynamic Environments

In recent years, there has been a growing interest in addressing dynamic optimization problems (DOPs) using evolutionary algorithms (EAs). Several approaches have been developed for EAs to increase the diversity of the population and enhance the performance of the algorithm for DOPs. Among these approaches, immigrants schemes have been found beneficial for EAs for DOPs. In this paper, random, elitism-based, and hybrid immigrants schemes are applied to ant colony optimization (ACO) for the dynamic travelling salesman problem (DTSP). Three ACO algorithms are proposed and compared with existing ACO approaches on a series of test DTSPs. The experimental results show that random immigrants are beneficial for ACO in fast changing environments, whereas elitism-based immigrants are beneficial for ACO in slowly changing environments. The ACO algorithm with the hybrid immigrants scheme attempts to combine the merits of the random immigrants and elitism-based immigrants schemes. Moreover, the results show that the proposed algorithms outperform compared approaches in almost all dynamic test cases and that immigrant schemes efficiently improve the performance of ACO algorithms in dynamic environments.