Distributed optimal power flow of AC/DC distribution networks with integer variables

Abstract

Due to the nature of integer variables, achieving multi-period optimal power flow (OPF) for AC/DC distribution networks (DNs) with integer variables via distributed algorithm presents significant technical challenges. These challenges primarily arise from two aspects. First, the integer variables lead to the traditional distributed algorithms failing to converge or converging to locally optimal solutions. Second, the binary variables related to topology switching cannot be relaxed to continuous variables, making existing distributed algorithms inapplicable. To address these problems, we propose the following solutions: (1) the primal problem is decomposed into a master problem and multiple subproblems using the generalized Benders decomposition (GBD) algorithm, and an iterative process is defined to obtain the optimal solution; (2) employing the branch-and-bound method (BBM) to avoid uncomputable cutting planes due to integer variables in the subproblem; and (3) using two acceleration techniques to improve efficiency by warming up the BBM and reducing parameter transfer between the master and subproblems. The effectiveness and accuracy of the proposed method are proven by simulation results of a test system, showing that the algorithm converges to the optimal solution in finite time. Additionally, the distributed algorithm provides a practical solution for operators and enhances privacy.