Rock-Climbing Group: An Innovative Meta-Heuristic Approach for Efficiently Tackling Optimization Problems

This paper introduces the Rock-Climbing Group (RCG) algorithm to tackle both constrained and global optimization problems. Contributions of this work include introducing a novel swarm intelligence algorithm inspired by rock climbing, which is both user-friendly and avoids local optima. The algorithm can be implemented in parallel to enhance efficiency and has demonstrated robustness and flexibility across various optimization scenarios. The introduced algorithm is straightforward and requires no control parameters. The RCG optimizer undergoes testing with an extensive collection of multimodal benchmark functions. Four traditional (CSA, DE, PSO and SCA) and four recently reported metaheuristic algorithms (GLA, MOA, SO and WSO) have been chosen from the literature, and comparative results are provided. The results show that RCG demonstrates superior convergence in ۱۸ out of ۲۰ test functions. Moreover, The RCG algorithm outperforms the other algorithms in terms of fitness for ۵۰ out of ۶۰ test functions. In terms of execution time, the RCG algorithm ranks sixth, surpassing WSO, SO, and MOA. Afterward, the RCG algorithm is used to solve the image segmentation problem and determine optimal threshold levels. Statistical tools show that RCG performs the best in ۴ out of ۹ test images.

کلیدواژه ها:

Rock Climbing Group ، Real World Problem ، Constrained problems ، Swarm Intelligence