Alexandria Engineering Journal · 2024 · 32 citations · 92 references
ArchitectureComputational EngineeringDesign OptimizationEvolutionary DesignOptimizationConstraint HandlingEvolutionary ComputationEvolutionary OptimizationGenetic AlgorithmsMetaheuristic OptimizationPopulation-based SearchAdaptive OptimizationArtificial Bee ColonyConvergenceEvolutionary AlgorithmsFirefly AlgorithmGlobal Search AlgorithmsPopulation Diversity Maintenance
The butterfly optimization algorithm (BOA) is a meta-heuristic algorithm that mimics foraging and mating behavior of butterflies. In order to alleviate the problems of slow convergence, local optimum and lack of population diversity of BOA, an enhanced adaptive butterfly optimization algorithm (EABOA) is proposed in this paper. First, a new adaptive fragrance model is designed, which provided a finer fragrance perception way and effectively enhanced the convergence speed and accuracy. Second, Lévy flight with high-frequency short-step jumping and low-frequency long-step walking is adopted to help the algorithm jump out of the local optimum. Third, the dimension learning-based hunting is employed to enhance information exchange by creating neighbors for each butterfly, thus improving the balance between local and global search and maintaining population diversity. In addition, the Fitness-Distance-Constraint (FDC) method is introduced to enhance constraint handling in EABOA (named FDC-EABOA). The proposed EABOA is compared with 8 well-known algorithms and 8 BOA variants in CEC 2022 test suite and the results were statistically analyzed using Friedman, Friedman aligned rank, Wilcoxon signed rank, Quade rank and multiple comparisons, analysis of variance (ANOVA) and range analysis. Finally, EABOA and FDC-EABOA are applied to seven engineering problems (parameter identification of photovoltaic module model, speed reducer design, tension/compression spring design, pressure vessel design, gear train design, welded beam design, SOPWM for 3-level inverters), and metrics such as Improvement Index (IF) and Mean Constraint Violation (MV) confirm that the proposed algorithms are satisfactory. Experimental results and statistical analysis show that the proposed algorithms outperform the comparison algorithms and demonstrate the strong potential for solving numerical optimization and engineering design problems.
92
Optimization by Simulated Annealing
Scott Kirkpatrick, C. D. Gelatt, M.P. Vecchi · Science · 1983 · 44.6K citations
Seyedali Mirjalili, Seyed Mohammad Mirjalili, Seyed Mohammad Mirjalili et al. · Advances in Engineering Software · 2014 · 18.4K citations · Full text
Joaquín Derrac, Salvador García, Daniel Molina et al. · Swarm and Evolutionary Computation · 2011 · 5.4K citations
Marco Dorigo, Mauro Birattari, Thomas Stützle · IEEE Computational Intelligence Magazine · 2006 · 5.2K citations
Fred Glover · INFORMS Journal on Computing · 1989 · 5K citations