Purpose Metaheuristic algorithms often suffer from premature convergence and poor population diversity when solving complex optimization issues. This study aims to develop an improved hybrid algorithm to balance global exploration and local exploitation, and adapt it to multi-objective constrained engineering optimization. Design/methodology/approach A Marsh Wren-Cuckoo Search algorithm is proposed by combining marsh wren decoy nest mechanism and cuckoo Levy flight strategy. Non-dominated sorting and crowding distance are adopted to construct its multi-objective version. Comparative experiments are carried out on standard benchmark functions and CEC-2022 test functions, with mainstream intelligent algorithms chosen as contrast models. Multiple indexes and statistical tests are used for performance assessment, and an I-beam design case is applied for practical verification. Findings The presented algorithm greatly restrains premature convergence and acquires superior convergence precision and stability. Statistical results prove its competitive performance. The multi-objective variant performs well on constraint engineering optimization tasks. Originality/value This work innovatively fuses two biological behaviors to maintain population diversity without losing convergence speed. It provides a reliable optimization tool for practical complex engineering design problems.
原文链接
分享
引用
