Evaluating Algorithm Efficiency in Large-Scale Dome Truss Optimization Under Frequency Constraints

Incorporating frequency constraints into the optimum design of large-scale truss dome structures is crucial for maintaining seismic resilience, as the natural frequencies must remain within specified ranges. In this work, seven metaheuristic algorithms—including three variants of the Fitness–Distance–Balance-based Adaptive Guided Differential Evolution (FDB-AGDE), the Cheetah Optimizer (CO), the Bonobo Optimizer (BO), the Flood Algorithm (FLA), and the Lung Performance Optimization (LPO) are applied to solve high-dimensional truss sizing problems under strict frequency limitations. Their convergence characteristics and solution quality are systematically compared across multiple dome configurations. Besides traditional measures of computational efficiency and final weight minimization, a suite of statistical analyses is conducted: the Wilcoxon rank-sum test to assess pairwise performance significance, the Friedman test to establish overall rank ordering, and Cohen’s test to quantify effect sizes. The results reveal that LPO, BO, CO, and the first variant of FDB-AGDE consistently produce lighter feasible designs with lower variability, whereas FLA and other variants of FDB-AGDE exhibit heavier structures or higher dispersion. The findings underscore the value of robust, well-tuned metaheuristics and rigorous statistical evaluation in structural optimization, offering clear guidance for seismic-focused designers seeking both lightweight solutions and reliable performance across repeated runs.