Multi-objective firefly algorithm with adaptive region division
Jia Zhao,Dandan Chen,Renbin Xiao,Juan Chen,Jeng-Shyang Pan,ZhiHua Cui,Hui Wang
DOI: https://doi.org/10.1016/j.asoc.2023.110796
2023-09-11
Applied Soft Computing Journal
Abstract:Aiming at the problems of single optimization strategy and poor comprehensive performance of MOFA , multi-objective firefly algorithm with adaptive region division is proposed in this paper. By leveraging the convergence index, our algorithm intelligently divides the dominant and non-dominant solution groups into three sub-regions, namely balance, exploration, and development areas, each with a distinct learning strategy that complements the strengths of fireflies. Specifically, fireflies in the balance area learn from global optimal particles with diversity to achieve a balanced exploration and development ability. Fireflies in the exploration area jointly learn from globally optimal particles with convergence and diversity, increasing the algorithm's likelihood of discovering Pareto optimal solutions. Lastly, fireflies in the development area rapidly converge under the guidance of the globally optimal particle of convergence, thus improving the algorithm's development ability. To further enhance the comprehensive optimization performance , we introduce a novel fusion index as an external archive update strategy that preserves solutions with superior convergence and diversity. Our experiments on 20 benchmark functions and a multi-objective optimization power flow example demonstrate that our algorithm outperforms other multi-objective optimization algorithms, highlighting its superior optimization performance.