A Dynamic Optimization Algorithm Based on Energy Level Collaboration Mechanism

Area of Science:

• Computer Science
• Artificial Intelligence
• Optimization Algorithms

Background:

• Multimodal optimization problems present challenges due to local optima, hindering conventional algorithms.
• Balancing global exploration and local exploitation is crucial for population-based optimizers.

Purpose of the Study:

• To propose a novel dynamic search framework optimization algorithm (DSF-ELC) for complex multimodal optimization.
• To enhance the adaptive exploration-exploitation balance in optimization algorithms.

Main Methods:

• Introduced a dynamic search framework optimization algorithm (DSF-ELC).
• Implemented a population dynamic reorganization strategy for adaptive particle migration.
• Developed a comprehensive learning strategy for dimension-wise learning from superior solutions.

Main Results:

• DSF-ELC demonstrated superior solution accuracy and stability on the CEC 2017 benchmark suite.
• Statistical tests (Wilcoxon signed-rank) and visualizations confirmed the algorithm's effectiveness.
• The proposed strategies achieved an adaptive exploration-exploitation balance.

Conclusions:

• DSF-ELC offers significant advantages for complex multimodal optimization problems.
• The algorithm shows broad application prospects in machine learning, engineering, and data science.