Implementing a dung beetle optimization algorithm enhanced with multi-strategy fusion techniques

Abstract

The dung beetle optimization algorithm possesses robust search and optimization capabilities. However, when it encounters complex optimization challenges, it struggles with limited accuracy, restricted global search ability, and suboptimal results from iterative optimization processes. To address these limitations, this study introduces a multi-strategy improved dung beetle optimization algorithm (CDBO). Initially, the golden sine method is implemented during the rolling phase to improve the algorithm’s capacity for both late area mining and early broad exploration; Then, in order to move the population closer to the ideal location during the foraging phase, the self-spiral method of the whale optimization algorithm is adopted. In the meanwhile, the present optimal location is arbitrarily perturbed during the stealing phase by introducing the flight of Levy strategy; Ultimately, the global optimal solution is modified using the dynamic t-distribution to enhance the algorithm’s capacity to eliminate the regional optimal solution. This study presents simulation tests with other intelligent optimisation algorithms on 23 test functions. The outcomes demonstrate that when the dimension is 30, the enhanced method performs optimally on at least 21 test functions. The modified method still earns the top score on 22 test functions and keeps its great search capabilities when the dimension is raised to 100. The enhanced approach is applied to address K-means clustering and engineering optimization problems to further assess its potential. The findings indicate that the improved method significantly boosts both the convergence rate and the accuracy of the optimization process.