AMOBH: Adaptive Multiobjective Black Hole Algorithm

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Evolutionary Computation

Background:

• Multiobjective optimization problems (MOPs) are common in science and engineering.
• Existing evolutionary algorithms face challenges in balancing convergence and diversity.
• The black hole algorithm provides a novel basis for optimization.

Purpose of the Study:

• To introduce the Adaptive Multiobjective Black Hole Algorithm (AMOBH).
• To enhance Pareto front convergence and diversity using a novel cell density assessment.
• To adapt evolutionary strategies using Shannon entropy.

Main Methods:

• Development of the Adaptive Multiobjective Black Hole Algorithm (AMOBH).
• Introduction of a cell density metric for solution evaluation.
• Mapping the Pareto front to a parallel cell coordinate system.
• Adaptive strategy adjustment using Shannon entropy.
• Fitness evaluation combining cell density and cell dominance.

Main Results:

• AMOBH demonstrates superior performance compared to SPEA-II, PESA-II, NSGA-II, and MOEA/D.
• The algorithm achieves a good balance between convergence and diversity.
• AMOBH shows improved convergence rate, population diversity, and convergence.
• Effective obtention of subpopulations across different Pareto regions.
• Competitive time complexity against state-of-the-art methods.

Conclusions:

• AMOBH offers a promising new approach for multiobjective evolutionary optimization.
• The cell density metric is effective in maintaining Pareto front quality.
• The adaptive framework enhances the algorithm's robustness and efficiency.