Akademik Veri Yönetim Sistemi
IEEE ACCESS, cilt.14, ss.66644-66673, 2026 (SCI-Expanded, Scopus)
Optimization algorithms are widely used to solve nonlinear engineering problems that are characterized by complex and constrained search spaces. In this study, the Overcurrent Relay Coordination (ORC) problem and a microgrid directional overcurrent relay coordination (DORC) case are examined using the Saplings Growing up Algorithm (SGuA) and its initialization-enhanced variants. Rather than introducing a fundamentally new metaheuristic paradigm, the main contribution of this study is a systematic analysis of how different sowing strategies affect convergence behavior, stability, and solution quality in both benchmark and engineering optimization settings. For this purpose, 20 classical benchmark functions, including scalable CEC functions and fixed-dimensional cases, are used to evaluate the algorithm under ten distinct initialization schemes. The results show that initialization has a measurable but problemdependent effect on performance. In the benchmark experiments, beta-init(2.5,2.5) and lhs-init showed the most consistent overall performance, with the effect of initialization becoming more pronounced as the dimensionality increased. The win–tie–loss analysis also indicated that beta-init(2.5,2.5) was the strongest overall strategy relative to the random-init reference. In the ORC case, lhsdesign produced the best mean performance over repeated runs, whereas Exprnd (0.5) yielded the best single-run result. In the microgrid DORC case, Lognrnd (0, 0.5) became the best average-performing sowing strategy within the SGuA2 framework, although leading competing optimizers such as CMA-ES, ABC, and GWO still achieved lower objective values. Overall, the findings suggest that initialization is not simply a preliminary step in SGuAbased optimization, but a meaningful design component whose effect depends on the structure and difficulty of the target problem.