Metaheuristic-based hybrid system for optimal reconfiguration of electrical grids leveraging distributed generation allocation

This paper presents a hybrid optimization framework for the reconfiguration of radial electrical distribution networks with integrated distributed generation allocation. The problem is formulated as a constrained combinatorial optimization model in which the objective function minimizes active power losses subject to power flow equations, voltage magnitude limits, branch capacity constraints, and strict radial topology requirements. Radial feasibility is ensured through Prim’s constructive heuristic, which generates spanning-tree configurations, while solution refinement is performed using the Variable Neighborhood Descent metaheuristic to systematically explore first-degree topological exchanges. Steady-state analysis is conducted using the Python-Pandapower framework, enabling detailed modeling of distributed generation units and accurate evaluation of loss variation and voltage profile sensitivity. The proposed method was assessed on 136- and 415-bus test systems, and benchmarked against population-based reconfiguration metaheuristics from the literature. The results indicate superior loss reduction performance with competitive computational effort, while preserving feasibility at every iteration. The allocation of two distributed generation units produced active power loss reductions of up to 75 percent under nominal loading conditions considering a worst-case scenario. The source code and input data is provided as supplementary material to ensure reproducibility.