Elite Bernoulli-based mutated dung beetle algorithm for global complex problems and parameter estimation of solar...

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Metaheuristics

Background:

• The Dung Beetle Optimization (DBO) algorithm is a metaheuristic known for simplicity but suffers from slow convergence and local optima stagnation.
• High-dimensional and non-convex problems challenge DBO's exploration-exploitation balance.

Purpose of the Study:

• To introduce a novel enhanced variant, Elite Bernoulli-based Mutated Dung Beetle Optimizer with Local Escaping Operator (EBMLO-DBO).
• To improve convergence speed, search capability, and robustness of the DBO algorithm for complex optimization problems.

Main Methods:

• Integration of Bernoulli map-based initialization for enhanced population diversity.
• Application of Morlet Wavelet mutation for adaptive local refinements and escaping local optima.
• Inclusion of elite guidance and a local escaping operator (LEO) for accelerated convergence and refined exploitation.

Main Results:

• EBMLO-DBO achieved superior performance on CEC2017 and CEC2022 benchmark suites, outperforming eleven state-of-the-art algorithms.
• Achieved first rank in 50% of CEC2022 functions and demonstrated high accuracy in photovoltaic parameter estimation.
• Statistical analysis confirmed EBMLO-DBO's significant superiority over compared algorithms.

Conclusions:

• EBMLO-DBO offers significantly improved search performance and solution quality compared to the original DBO.
• The proposed enhancements effectively address DBO's limitations in convergence and robustness.
• EBMLO-DBO demonstrates strong potential for solving complex optimization problems and real-world applications.