Multi-population Evolutionary and Swarm Intelligence Dynamic Optimization Algorithms: A Survey

Multi-population evolutionary and swarm intelligence dynamic optimization algorithms are the most flexible and effective methods for solving dynamic optimization problems. In a dynamic optimization problem, the search space is affected by environmental changes over time. In multi-population evolutionary and swarm intelligence dynamic optimization algorithms, the number of sub-populations is a parameter determined either by the user or adaptively. The use of multiple sub-populations enables these methods to efficiently track the moving optimum. These methods are capable of gathering historical knowledge about the search space, which is used to effectively react to changes and provide a warmed-up start for the algorithm in new environments. In this chapter, the components of multi-population algorithms are classified to the ones that are used for sub-population formation, management of computational resources, transmission of information from previous environments, and handling diversity loss. Based on this classification, researchers can have a better understanding of how these components make evolutionary and swarm intelligence algorithms capable of addressing the challenges of dynamic optimization problems.