Multi-strategy driven crow search algorithm for global optimization and engineering problems

doi:10.1007/s10586-026-06543-6

Multi-strategy driven crow search algorithm for global optimization and engineering problems

2026-09-16

0

PRE

AI

Muhammed Kıvanc Kurnaz

DOI：10.1007/s10586-026-06543-6

Original

Original request for help

Save

Abstract

En 中文

The Crow Search Algorithm (CSA) is a swarm-based method inspired by and mimicking the foraging habits of crows. Although CSA has a strong space search strategy, its use of random individuals and random probabilities in the exploration phase and the simplicity of its search mode prevent it from finding the global best solution when solving complex problems. This paper proposes a multi-strategy-driven CSA to overcome the shortcomings of the standard CSA method, which is named MSD-CSA. A new hybrid updating strategy consisting of the guiding strategy of the Jellyfish Search Optimizer and refraction-based learning is proposed to improve the exploration phase of CSA. Furthermore, the spiral-based motion strategy enhances the efficacy of the exploitation phase. The effectiveness of the proposed algorithm is verified by comparing it with the original CSA, six other original algorithms, namely PSO, GWO, MFO, WOA, MVO, and TLBO, and three improved algorithms, namely eKGSA, EJAYA, and CCSA, on eleven well-known unimodal and nine multimodal benchmark functions. To analyze the performance of the proposed method on test functions, a series of statistical tests was performed, and convergence curves were analyzed to examine its behavior on problems with different characteristics. Furthermore, the robustness of the method was tested by considering issues of various dimensions. The performance of the method was analyzed on four engineering design problems involving gear train design, multi-disk clutch brake design problem, gas transmission system design problem, and reactor network design, with other optimization methods. Experimental results showed the superiority of the MSD-CSA method over other methods, achieving the best results for 63% of the applied unimodal and 89% of the applied multimodal benchmark functions. Furthermore, a robust control model for quadcopter altitude control was designed based on the proposed MSD-CSA method. In the analyses, the MSD-CSA method reached the reference altitude in a short time, achieved an overshoot of 0.0489, and an error rate of 6.05e-04. Analyses on different problems show that the MSD-CSA method provides a good balance between exploration and exploitation phases, making it effective in local and global search.

Keywords:

Crow Search Algorithm

Spiral-based Strategy

Refraction-Based Learning

Multi-strategy-driven Crow Search Algorithm

Quadcopter control

Engineering design optimization

Multi-strategy driven crow search algorithm for global optimization and engineering problems

Abstract

Journal

Organization

Cited Papers