Optimal Power Flow Based on Area Partitioning Method for Power Grids

The power system is gradually orienting to large-scale distributed computing, and the trend of area interconnection is becoming more and more obvious. It is still a classic prob-lem to calculate power flow by using distributed optimization. Compared with traditional centralized power flow calculation, it has obvious advantages with fast computing speed and less information between areas. In this paper, we define the boundary voltage coupling equation with auxiliary variables, to solve the non-convex subproblem in the iterative process of ADMM, and match the optimal area partitioning for ADMM by proposing the optimal spectral clustering (OSC) method. Specifically, the OSC scheme takes the Jacobian matrix, admittance matrix, and voltage pivot vector as the the joint adjacency matrix to measure the similarity of spectral clustering algorithm into consideration, which is obviously better than the traditional spectral clustering method considering admittance matrix only. Combining with ADMM algorithm, IEEE-118 bus system is used as a test example to perform the distributed calculation after area partitioning, the experimental results verify the good performance of proposed spectral clustering method in distributed power grid.