Abstract
Enterprise Development Optimizer (EDO) is a meta-heuristic algorithm inspired by the enterprise development process with strong global search capability. However, the analysis of the EDO algorithm shows that it suffers from the defects of rapidly decreasing population diversity and weak exploitation ability when dealing with complex optimization problems, while its algorithmic structure has room for further enhancement in the optimization process. In order to solve these challenges, this paper proposes a multi-strategy enterprise development optimizer called MSEDO based on basic EDO. A leader-based covariance learning strategy is proposed, aiming to strengthen the quality of search agents and alleviate the weak population diversity of the EDO algorithm in the later search stage through the guiding role of the dominant group and the modifying role of the leader. To dynamically improve the local exploitation capability of the EDO algorithm, a fitness and distance-based leader selection strategy is proposed. In addition, the structure of EDO algorithm is reconstructed and a diversity-based population restart strategy is presented. The strategy is utilized to assist the population to jump out of the local optimum when the population is stuck in search stagnation. Ablation experiments verify the effectiveness of the strategies of the MSEDO algorithm. The performance of the MSEDO algorithm is confirmed by comparing it with five different types of basic and improved metaheuristic algorithms. The experimental results of CEC2017 and CEC2022 show that MSEDO is effective in escaping from local optimums with its favorable exploitation and exploration capabilities. The experimental results of ten engineering constrained problems show that MSEDO has the ability to competently solve real-world complex optimization problems.