Dung beetle optimizer based on mean fitness distance balance and multi-strategy fusion for solving practical engineering problems

As a swarm intelligence algorithm, Dung beetle optimizer (DBO) was inspired by the behavior pattern of dung beetles for survival. This paper presents a dung beetle optimizer based on mean fitness distance balance and multi-strategy fusion (MMDBO), which addresses the slow convergence and weak global search of the original DBO. MMDBO improves performance by incorporating a cosine similarity strategy for position updates, enhancing convergence speed and diversity. The MFDB strategy is then applied to balance global exploration with local exploitation. Additionally, the Hypotrochoid and Levy flight strategies are used to enhance search ability and solution quality, while a non-uniform mutation strategy is introduced to avoid local optima. To comprehensively evaluate the optimizer performance of MMDBO, experiments were conducted on the IEEE CEC2017 and CEC2022 benchmark sets, comparing it with 13 other population-based optimizer algorithms. The experimental results show that MMDBO achieves Friedman mean rankings of 1.21, 1.52, and 1.52 for the 30-dimensional, 50-dimensional, and 100-dimensional problems on the CEC2017 benchmark set, respectively, and a ranking of 1.5 for the 20-dimensional problem on the CEC2022 benchmark set. These results indicate that MMDBO consistently outperforms most algorithms and provides accurate and reliable optimizer solutions. Additionally, MMDBO's practicality is demonstrated through five real-world constrained engineering design challenges, including a wind farm layout optimizer problem and a magnesium alloy constitutive model parameter identification problem, further validating its broad applicability in real engineering problems. The results of the study indicate that MMDBO possesses excellent optimizer capacity and broad application potential.