A Hybrid Nonlinear Greater Cane Rat Algorithm with Teaching-Learning-Based Optimization for Global Optimization and...

Area of Science:

• Computational Intelligence
• Swarm Intelligence Algorithms
• Optimization Techniques

Background:

• The Greater Cane Rat Algorithm (GCRA) is a swarm intelligence paradigm inspired by GCR survival behaviors.
• The original GCRA exhibits limitations in high-dimensional problems, including premature convergence and local optima stagnation.
• These limitations restrict the practical application of GCRA in complex engineering optimization.

Purpose of the Study:

• To introduce an enhanced hybrid variant of the GCRA, termed TLGCRA, by integrating Teaching-and-Learning-Based Optimization (TLBO).
• To address the inherent defects of the original GCRA, improving its performance in complex and high-dimensional optimization scenarios.
• To enhance convergence speed, solution precision, and algorithmic robustness for engineering applications.

Main Methods:

• Integration of TLBO's two-stage teacher-student learning mechanism into the GCRA framework.
• Implementation of an adaptive parameter tuning strategy to balance exploration and exploitation.
• Extensive computational simulations on 23 benchmark functions and 6 constrained engineering design problems.
• Benchmarking against canonical GCRA, LPSO, and ten other metaheuristic approaches.

Main Results:

• The TLGCRA demonstrated significant performance advantages over existing algorithms in convergence velocity, solution precision, and resilience.
• Marked improvement in optimal solution precision for complex multimodal functions was observed.
• Near-zero standard deviation across multiple runs in engineering cases confirmed TLGCRA's excellent stability.
• Statistical tests (Friedman, Wilcoxon signed-rank) validated the algorithm's robust optimization efficacy.

Conclusions:

• The proposed TLGCRA effectively overcomes the limitations of the original GCRA, offering enhanced environmental adaptability and comprehensive optimization performance.
• The hybrid algorithm achieves faster convergence, higher computational accuracy, and outstanding stability and robustness.
• TLGCRA provides a viable and effective framework for tackling intricate constrained engineering optimization challenges.