.png){kind=icon}
.png){kind=icon}
Return
Icicle algorithm: a novel physics-based metaheuristic algorithm
.png){kind=icon}
PRE
AI
DOI：10.1007/s10586-026-06331-2.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
In recent years, metaheuristic algorithms have achieved remarkable success across various fields of optimization; however, many still suffer from premature convergence, imbalanced exploration and exploitation, and sensitivity to parameter tuning. Nevertheless, to overcome these limitations, this paper proposes a novel physics-based metaheuristic algorithm named the Icicle Algorithm (IA). Inspired by the natural process of icicle formation, IA introduces a simple yet adaptive mechanism without additional control parameters, enabling a dynamic balance between global exploration and local exploitation through four physically inspired stages: melting, movement, solidification, and falling. To evaluate the performance of IA, comprehensive experimental evaluations were conducted on two benchmark test suites from the IEEE Congress on Evolutionary Computation (CEC-2017 and CEC-2022). Quantitative analysis based on Friedman mean rank results shows that IA ranks 2.34, 1.31, 1.24, and 1.55 on the CEC-2017 test suite with 10, 30, 50, and 100 dimensions, respectively, and achieves a mean rank of 2.08 on the CEC-2022 test suite with 10 dimensions. These results indicate that IA achieves competitive or superior performance compared with twelve classical and recently developed metaheuristic algorithms. Furthermore, IA is applied to five classical engineering optimization problems, including the pressure vessel, welded beam, tension/compression spring, speed reducer, and three-bar truss design problems, demonstrating its robustness and practical applicability. The results confirm that IA not only exhibits strong optimization performance on benchmark suites but also provides high-quality solutions in real-world engineering scenarios. The source code of IA is publicly available at: https://github.com/Gang1024/Icicle-Algorithm .
Keywords:
Optimization
Metaheuristic algorithms
Nature-inspired
Icicle algorithm
Engineering design optimization problems
Journal
IF:
2.9
Papers:
176
Citations:
1.1K