A Multi-Strategy Improved Dung Beetle Optimizer for High-Dimensional Optimization and Engineering Applications

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Metaheuristics

Background:

• The standard Dung Beetle Optimizer (DBO) exhibits limitations in high-dimensional optimization, including slow convergence, local optima stagnation, and reduced population diversity.
• Addressing these inherent defects is crucial for effective problem-solving in complex computational tasks.

Purpose of the Study:

• To propose a novel, multi-strategy hybrid improved DBO variant, named SWan Dung Beetle Optimizer (SWDBO).
• To enhance the DBO's capability in tackling high-dimensional numerical and constrained engineering optimization problems.

Main Methods:

• Incorporation of an adaptive population proportion strategy to dynamically manage beetle populations for balanced exploration and exploitation.
• Integration of Whale Optimization Algorithm (WOA) mechanisms (bubble-net encircling, spiral predation) into rolling beetle position updates.
• Introduction of a modified seagull optimization operator with Lévy random perturbation for thief beetle position updates to improve escape from local optima.

Main Results:

• Numerical experiments on CEC2017 and CEC2020 benchmark functions demonstrated the SWDBO's effectiveness.
• Validation on three engineering tasks (three-bar truss, ten-bar truss, cantilever beam design) showed lighter structural mass with satisfied constraints.
• Wilcoxon rank-sum tests confirmed statistically significant performance improvements over competing optimizers.

Conclusions:

• The proposed multi-strategy improvement framework significantly enhances optimization performance.
• SWDBO effectively balances global exploration and local exploitation, outperforming traditional DBO and other optimizers.
• SWDBO is a promising approach for high-dimensional numerical and constrained engineering optimization challenges.