A supercomputing method for large-scale optimization: a feedback biogeography-based optimization with steepest descent method

Zhang, Ziyu; Gao, Yuelin; Guo, Eryang

doi:10.1007/s11227-022-04644-8

Cited by 1 publication

(1 citation statement)

References 71 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In addition, BBO relies only on the replacement of variables between different habitats to search for new solutions, which cannot balance the exploration and exploitation well and has a slow convergence speed. These are the main reasons why BBO cannot effectively solve high-dimensional numerical problems [38].…”

Section: Standard Bbomentioning

confidence: 99%

A Comprehensive Multi-Strategy Enhanced Biogeography-Based Optimization Algorithm for High-Dimensional Optimization and Engineering Design Problems

Gao,

Li,

Gao

et al. 2024

Mathematics

Self Cite

View full text Add to dashboard Cite

The biogeography-based optimization (BBO) algorithm is known for its simplicity and low computational overhead, but it often struggles with falling into local optima and slow convergence speed. Against this background, this work presents a multi-strategy enhanced BBO variant, named MSBBO. Firstly, the example chasing strategy is proposed to eliminate the destruction of the inferior solutions to superior solutions. Secondly, the heuristic crossover strategy is designed to enhance the search ability of the population. Finally, the prey search–attack strategy is used to balance the exploration and exploitation. To verify the performance of MSBBO, we compare it with standard BBO, seven BBO variants (PRBBO, BBOSB, HGBBO, FABBO, BLEHO, MPBBO and BBOIMAM) and seven meta-heuristic algorithms (GWO, WOA, SSA, ChOA, MPA, GJO and BWO) on multiple dimensions of 24 benchmark functions. It concludes that MSBBO significantly outperforms all competitors both on convergence accuracy, speed and stability, and MSBBO basically converges to the same results on 10,000 dimensions as on 1000 dimensions. Further, MSBBO is applied to six real-world engineering design problems. The experimental results show that our work is still more competitive than other latest optimization techniques (COA, EDO, OMA, SHO and SCSO) on constrained optimization problems.

show abstract

Section: Standard Bbomentioning

confidence: 99%