Flexible mesh-shaped distribution network topology with adaptive geographical layered-clustered probabilistic planning

With the large-scale integration of distributed renewable energy, traditional distribution network topologies lack the flexibility to fully exploit wide-area source-load complementarity and flexible resources, limiting the efficient and secure accommodation of distributed generation (DG). To address this, we propose a mesh-shaped distribution network topology with a three-terminal soft open point (SOP) as the core device, balancing existing grid features with retrofit complexity. Given the high-dimensional, complex MILP resulting from planning under strong stochasticity, we introduce a novel geographical layered-clustered planning method. This method, utilizing an improved fast-unfolding algorithm, considers spatio-temporal source-load distribution, customer reliability, and line-construction needs during clustering, effectively decomposing the planning problem into parallelizable subproblems to improve efficiency. A real-world case study validates the proposed method, identifying the most suitable application scenarios for the mesh-shaped topology. Compared to conventional feeder-partition-based planning, our method accommodates current target structures’ construction requirements while saving over 80% of computation time with only a 1.71% loss in optimality.