Genetic Algorithms with Adaptive Immigrants for Dynamic Environments

One approach integrated with genetic algorithms (GAs) to address dynamic optimization problems (DOPs) is to maintain diversity of the population via introducing immigrants. Many immigrants schemes have been proposed that differ on the way new individuals are generated, e. g., mutating the best individual of the previous environment to generate elitism-based immigrants. This paper examines the performance of elitism-based immigrants GA (EIGA) with different immigrant mutation probabilities and proposes an adaptive mechanism that tends to improve the performance in DOPs. Our experimental study shows that the proposed adaptive immigrants GA outperforms EIGA in almost all dynamic test cases and avoids the tedious work of fine-tuning the immigrant mutation probability parameter.