An efficient feasible region search approach for scalable power distribution system reconfiguration

This paper presents a novel algorithm based on a feasible region search strategy applied to the Power Distribution System Reconfiguration (PDSR) problem. The proposed method utilizes specialized in-memory data structures, specifically hash maps, to accelerate the optimization process by strictly confining the search to the feasible region, ensuring both radiality and connectivity at every iteration. Our central hypothesis posits that the high computational cost of traditional algorithms in the literature stems from the time wasted evaluating infeasible candidate solutions through expensive power flow calculations. By ensuring that only feasible candidates are generated, the proposed algorithm eliminates unnecessary power flow executions. Simulations conducted on traditional test systems demonstrate that the proposed method is a highly promising solution; for the largest system tested (417-bus), a computational speedup of approximately 15 times was achieved compared to the best execution times reported in the literature. Furthermore, the results indicate that the execution time scales linearly rather than exponentially with the system size, facilitating its application in real-time operation of large-scale distribution networks.