Fitness Distance Balanced Starfish Optimization for Benchmark and Engineering Design Problems.

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Biomimetic Computing

Background:

• Biomimetic optimizers require balancing diversity preservation and selection pressure for complex engineering tasks.
• Existing algorithms like the Starfish Optimization Algorithm (SFOA) can be improved by incorporating fitness-distance awareness.

Purpose of the Study:

• To enhance the Starfish Optimization Algorithm (SFOA) using fitness-distance-aware selection control.
• To introduce two variants, FDBSFOA and dFDBSFOA, to improve diversity and convergence in optimization.
• To evaluate the effectiveness of these enhanced algorithms on benchmark and engineering problems.

Main Methods:

• Developed Fitness-Distance Balance Starfish Optimization Algorithm (FDBSFOA) and its dynamic variant (dFDBSFOA).
• Guided candidate selection using solution quality and spatial diversity relative to the best solution.
• Evaluated algorithms on IEEE CEC2017, CEC2020, CEC2022 benchmark suites and constrained engineering design problems.

Main Results:

• The proposed variants significantly improve the robustness and search efficiency of the baseline SFOA.
• dFDBSFOA demonstrated the most consistent overall performance across tested problems.
• The enhanced algorithms showed controlled and interpretable computational overhead.

Conclusions:

• Diversity-aware selection is an effective principle for strengthening biomimetic optimization frameworks.
• FDBSFOA and dFDBSFOA offer improved performance for continuous, single-objective optimization problems.
• Future work includes extending the strategy to dynamic, noisy, and multi-objective optimization settings.