The cheetah optimizer: a nature-inspired metaheuristic algorithm for large-scale optimization problems

TLDR

Cheetahs employ three primary hunting strategies—searching, sitting‑and‑waiting, and attacking—to capture prey. This study introduces the cheetah optimizer (CO), a nature‑inspired algorithm modeled after these hunting tactics. CO incorporates the three core strategies and an additional leave‑the‑prey‑and‑return tactic to enhance population diversity, convergence, and robustness, and is evaluated on 14 shifted‑rotated CEC‑2005 functions, the CEC‑2010 and CEC‑2013 large‑scale benchmarks, and the economic load dispatch problem. Across all tests, CO consistently outperforms conventional and state‑of‑the‑art algorithms, demonstrating superior performance on large‑scale and complex optimization tasks. The CO algorithm’s source code is publicly available at https://www.optim-app.com/projects/co.

Abstract

Abstract Motivated by the hunting strategies of cheetahs, this paper proposes a nature-inspired algorithm called the cheetah optimizer (CO). Cheetahs generally utilize three main strategies for hunting prey, i.e., searching, sitting-and-waiting, and attacking. These strategies are adopted in this work. Additionally, the leave the pray and go back home strategy is also incorporated in the hunting process to improve the proposed framework's population diversification, convergence performance, and robustness. We perform intensive testing over 14 shifted-rotated CEC-2005 benchmark functions to evaluate the performance of the proposed CO in comparison to state-of-the-art algorithms. Moreover, to test the power of the proposed CO algorithm over large-scale optimization problems, the CEC2010 and the CEC2013 benchmarks are considered. The proposed algorithm is also tested in solving one of the well-known and complex engineering problems, i.e., the economic load dispatch problem. For all considered problems, the results are shown to outperform those obtained using other conventional and improved algorithms. The simulation results demonstrate that the CO algorithm can successfully solve large-scale and challenging optimization problems and offers a significant advantage over different standards and improved and hybrid existing algorithms. Note that the source code of the CO algorithm is publicly available at https://www.optim-app.com/projects/co .