Crowding-based Differential Evolution with Self-Adaptive Control Parameters for Dynamic Environments

Many of optimization problems possess dynamic characteristics. Introducing strategies to maintain population diversity plays a key role in Dynamic Optimization Problems (DOPs), thus making the balance of exploration ability and the exploitation ability of the proposed algorithm. In this study, a crowding based differential evolution, combined with self-adaptive control parameters (CaDE)was proposed. Our proposed method adopts crowding with the least individual within a predefined neighborhood radius, and the control parameters are self-adaptive during the evolution process. For comparative analysis, several schemes incorporated with differential evolution are tested on a widely used moving peaks benchmark (MPB). Performance of the proposed algorithm is validated on the selected benchmark function in each circumstance.