Multi-population dynamic grey wolf optimizer based on dimension learning and Laplace Mutation for global optimization

Metaheuristic algorithms are highly popular in the field of optimization because of their gradient-free nature and strong applicability. Grey Wolf Optimizer (GWO for short) performs well in optimization problems in different fields due to its fast convergence speed and strong exploitation ability. However, GWO still has shortcomings in terms of population diversity, balance between exploration and exploitation, and convergence during the optimization process. To address these shortcomings, this article proposes a multi-population dynamic grey wolf optimizer based on dimension learning and Laplace mutation (DLMDGWO). Firstly, DLMDGWO is used in the initial stage to obtain a high-quality and uniform initial population by controlling the dynamic boundaries so that the population is distributed in Euclidean distances around the origin of the search space and perturbed by diversity through the logistics map. Moreover, utilizing Multi-population Dynamic Strategy to improve the algorithm mechanism of grey wolves and proposing Double Laplace Mutation to enrich the update strategy of grey wolves. In addition, Multi-strategy Dimension Learning is utilized to optimize the dynamic structure of the population and enhance the diversity of the grey wolf in the solving process. In order to investigate the effectiveness of the proposed DLMDGWO, it has been tested on the standard benchmark functions in IEEE CEC 2017 and IEEE CEC 2022. In addition, DLMDGWO was also used to solve eight practical engineering optimization problems. The results of the analysis and comparison with other improved GWO indicate that the proposed DLMDGWO has better on search efficiency, solution accuracy, and convergence speed in executing global optimization problems.