A Novel Hybrid Improved RIME Algorithm for Global Optimization Problems.

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Metaheuristic Computing

Background:

• The RIME algorithm, a physical-based meta-heuristic, excels at global optimization but suffers from limited exploration, slow convergence, and exploration-exploitation asymmetry.
• Engineering applications often require robust optimization methods capable of handling complex global problems.

Purpose of the Study:

• To introduce Hybrid Estimation Rime-ice Optimization (HERIME), an enhanced meta-heuristic algorithm.
• To address the limitations of the original RIME algorithm, specifically improving exploration, convergence speed, and the balance between exploration and exploitation.

Main Methods:

• Incorporation of a probabilistic model-based sampling approach from the Estimation of Distribution Algorithm to improve population quality and global exploration.
• Implementation of a roulette-based fitness distance balanced selection strategy to enhance the hard-rime phase and balance optimization phases.
• Validation using 41 functions from the IEEE CEC2017 and IEEE CEC2022 test suites.

Main Results:

• HERIME significantly outperformed classical, recent, and advanced metaheuristic algorithms in optimization accuracy, convergence, and stability.
• Statistical tests (Friedman, Wilcoxon rank sum) confirmed the superior performance of HERIME.
• Ablation experiments validated the effectiveness of the individual strategies integrated into HERIME.

Conclusions:

• HERIME offers improved search efficiency and optimization accuracy compared to existing methods.
• The proposed algorithm is highly effective for tackling global optimization problems in engineering and other fields.