Abstract
Distribution network reconfiguration requires connected radial operation to be maintained while searching a large discrete space, a challenge intensified by distributed generation. This paper proposes the Discrete Adaptive Hybrid Seagull Optimization Algorithm, which reformulates the continuous migration and attack operators of the Seagull Optimization Algorithm on Fundamental Loop decision sets. Rather than chaining migration and attack in every iteration as in the original continuous update, an iteration-dependent stochastic schedule selects exactly one discrete operator for the entire population. A decaying control factor regulates migration intensity, and every candidate is evaluated through steady-state alternating-current power flow using OpenDSS. Across 100 independent runs, the proposed method reaches the 33-bus best-known loss of 139.55 kW in every run and yields a mean loss of 279.91 kW on the 135-bus system. A seven-variant ablation identifies migration as the principal performance driver and shows an additional robustness contribution from attack and stochastic phase selection on the larger network, while not supporting a claim that the fixed attack step is uniquely optimal. Fixed-siting DG sweeps with 100 runs at each penetration level identify the lowest mean loss at 71% penetration in the 33-bus system. For the 135-bus system, the results indicate a statistically overlapping low-loss region at 83–84% penetration. All 20,000 penetration-study solutions satisfy the implemented convergence, voltage, connectivity, and radiality checks. The evidence supports the method as a competitive stochastic optimizer under static, deterministic operating conditions and motivates future uncertainty-aware and multi-period extensions.