An Enhanced Slime Mould Algorithm Based on Best-Worst Management for Numerical Optimization Problems.

Area of Science:

• Swarm Intelligence
• Computational Intelligence
• Optimization Algorithms

Background:

• The Slime Mould Algorithm (SMA) is a popular swarm intelligence method.
• Existing SMA versions face limitations, including slow convergence and local optima.
• The 'no free lunch' theorem highlights the need for algorithm specialization and improvement.

Purpose of the Study:

• To propose a new variant of the Slime Mould Algorithm (SMA), named BWSMA.
• To enhance the SMA's convergence speed, population quality, and ability to escape local optima.
• To validate the effectiveness and robustness of the BWSMA through comprehensive experiments.

Main Methods:

• Integration of three novel mechanisms into the SMA: adaptive greedy, best-worst management, and stagnant replacement.
• Extensive experimental validation using the CEC2018 and CEC2022 benchmark test suites.
• Comparative analysis against three derived algorithms, eight SMA variants, and eight other improved algorithms.
• Statistical analysis using Wilcoxon rank-sum, Friedman, and Nemenyi tests.
• Application to two structural optimization problems to assess real-world applicability.

Main Results:

• The BWSMA significantly outperformed all compared algorithms across various test suites.
• BWSMA achieved superior average rankings compared to SMA variants and other improved algorithms.
• Statistical tests confirmed the significant performance advantage of the BWSMA.
• The algorithm demonstrated strong applicability in solving structural optimization problems.

Conclusions:

• The proposed BWSMA is a highly effective and robust optimization algorithm.
• The integrated mechanisms successfully address the shortcomings of the original SMA.
• BWSMA offers excellent search accuracy and is a promising advancement in swarm intelligence.