A three-phase sheep optimization algorithm for numerical and engineering optimization problems

doi:10.1016/j.eswa.2024.123338

2024-08-01

PRE

DOI：10.1016/j.eswa.2024.123338

原文链接

原文求助

摘要

En 中文

Swarm intelligence represents an artificial intelligence discipline, founded upon the principles governing the conduct of animal societies. The sheep optimization algorithm (SO), a novel contribution to the field, emulates three distinct sheep behaviors, namely bellwether guidance (BGD), sheep interaction (SIA), and shepherd dog supervision (SDS). To augment its performance, we devise an enhanced method known as the improved sheep optimization algorithm (ISO), drawing upon the salient features of these three strategies. In particular, we introduce the gradient descent strategy (GDS) to expedite the convergence rate of the BGD process. For the SIA procedure, we employ a dimension by dimension improvement (DDI) technique, enabling the exploration of a broader solution space. In the SDS stage, we adopt an effective variable neighborhood search (VNS) to strike a balance between exploration and exploitation. Furthermore, ISO assumes the role of a versatile algorithm framework, allowing for the customization and adaptation of the aforementioned strategies to cater to specific solutions. To evaluate the efficacy of ISO, we test it using the CEC 2017 test suite. Comparative analyses are conducted against state-of-the-art solvers in the CEC competition. Experimental results convincingly demonstrate that ISO outperforms other state-of-the-art algorithms in terms of solution quality, convergence speed, and solution stability. Lastly, ISO effectively showcases its advancement by successfully tackling two real -world engineering problems.

Keyword:

Gradient descent strategy

Sheep optimization

Dimension by dimension improvement

Variable neighborhood search