Artificial bee Colony Algorithm with Dynamic Dimension Search

Aiming at the shortcomings of the standard Artificial Bee Colony (ABC) algorithm due to weak exploitation capability and slow convergence speed caused by one-dimension search, an ABC algorithm with dynamic dimension search (DDSABC) is proposed. Firstly, a dynamic selection mechanism for dimension search strategy is designed, where one-dimension search is used in the early stage of search to exploit globally diverse individuals, and multi-dimension search is used in the late stage to explore convergent individuals; an elite moderately guided dimension selection mechanism is designed for multi-dimension search, where half of the dimensions are selected under the guidance of the elite mean to accelerate the evolution of potential dimensions, and the other half are selected by a randomized way to maintain a moderate randomness; in addition, suitable search equations are selected for the different characteristics of the two dimension search strategies to ensure the balance between exploration and exploitation capabilities. Finally, on 30 complex problems from the CEC2014 test set, ablation experiments and comparative experiments with the standard ABC and its 5 variants are analyzed, and the results show that the proposed algorithm in this paper performs better in terms of solution accuracy, stability and convergence speed.