A Novel Hybrid Empirical Discrete Metaheuristic and Simplex Algorithm With Branch Separation for Optimal Photovoltaic Allocation in Radial Distribution Networks

ABSTRACT

Optimal placement and sizing of photovoltaic (PV) plants in radial distribution networks remains challenging due to the trade‐off between global search and local precision. This paper proposes a hybrid algorithm that combines a modified empirical discrete metaheuristic (EDM), the simplex complex method, and a novel branch separation division (DSR) strategy. The DSR leverages the radial topology for structured local searches, and a ‘window’ parameter prevents electrically adjacent plant placements. A multi‐objective function minimises system voltage deviation and energy losses. Compared to an EDM+coordinate search (EDM+CSM) benchmark on the IEEE 33‐bus and a 286‐bus Cuban network, the proposed method achieves significantly higher robustness, finding high quality near global optima in 58% vs. 9% of runs on the 33‐bus system and 15% vs. 1% on the 286‐bus system, while reducing average execution time by nearly 40 s on the larger network. A sensitivity analysis of the weight factor α shows that α ≈ 0.5 best balances loss reduction and voltage profile. For the Cuban system, marginal technical gains justify a maximum of three PV plants.