Salgotra Rohit, Singh Supreet, Verma Pooja, Abualigah Laith, Gandomi Amir H
Faculty of Physics and Applied Computer Science / Centre of Excellence in Artificial Intelligence, AGH University of Krakow, Kraków, Poland.
Faculty of Engineering & Information Technology, University of Technology Sydney, Sydney, Australia.
Sci Rep. 2025 Jun 4;15(1):19655. doi: 10.1038/s41598-025-01033-y.
Cuckoo Search (CS) is a popular algorithm used to solve numerous challenging problems. In the present work, a novel variant of CS is presented to eliminate its shortcomings. The proposed algorithm is hybridized with the naked mole rat algorithm (NMRA) to enhance the exploitative behavior of CS, and is called Mutated Adaptive Cuckoo Search Algorithm (MaCN). This new algorithm has self-adaptive properties and its key feature is to divide the solutions into multiple sections, which are often called sub-swarms. In addition, a bare-bones search mechanism is also added to enhance exploration. The use of adaptive inertia weights helps optimize the switching probability, an important CS parameter that helps to achieve a balanced operation. The proposed MaCN algorithm is tested on CEC 2005 and CEC 2014 benchmark problems. Comparative studies showed that MaCN delivers promising results in solving CEC competition benchmark problems compared to JADE, success history-based adaptive DE (SHADE), LSHADE-SPACMA and self-adaptive DE (SaDE), among others. In addition to numerical benchmarks, MaCN is used to solve the industrial engineering frame structure and a comparison with hybridization of particle swarm with passive congregation (PSOPC), shuffled frog leaping algorithm hybrid with invasive weed optimization (SFLAIWO), particle swarm ant colony optimization (PSOACO), early strategy with DE (ES-DE), and others show its superiority. In addition, the Wilcoxon rankum and the Freidmann test statistically prove the significance of the proposed MaCN algorithm. MaCN was found to score first rank for the benchmarks. The application of the MaCN algorithm to solve the design problems of the suggests that the best new results are obtained for all test cases.
布谷鸟搜索算法（CS）是一种常用于解决众多挑战性问题的流行算法。在当前工作中，提出了一种新型的CS变体以消除其缺点。所提出的算法与裸鼹鼠算法（NMRA）进行了杂交，以增强CS的开发行为，该算法被称为变异自适应布谷鸟搜索算法（MaCN）。这种新算法具有自适应特性，其关键特征是将解划分为多个部分，这些部分通常被称为子群。此外，还添加了一种骨架搜索机制以增强探索能力。自适应惯性权重的使用有助于优化切换概率，这是一个有助于实现平衡操作的重要CS参数。所提出的MaCN算法在CEC 2005和CEC 2014基准问题上进行了测试。比较研究表明，与JADE、基于成功历史的自适应差分进化算法（SHADE）、LSHADE - SPACMA和自适应差分进化算法（SaDE）等相比，MaCN在解决CEC竞赛基准问题方面取得了有前景的结果。除了数值基准测试外，MaCN还用于解决工业工程框架结构问题，并且与粒子群与被动聚集杂交算法（PSOPC）、混合洗牌蛙跳算法与入侵杂草优化算法（SFLAIWO）、粒子群蚁群优化算法（PSOACO）、早期策略与差分进化算法（ES - DE）等的比较显示了其优越性。此外，威尔科克森秩和检验与弗里德曼检验从统计学上证明了所提出的MaCN算法的显著性。发现MaCN在基准测试中排名第一。MaCN算法在解决设计问题中的应用表明，所有测试案例均获得了最佳新结果。
