Data‐driven approach for real‐time distribution network reconfiguration
Ziyang Yin, Xingquan Ji, Yumin Zhang, Qi Liu, Xingzhen Bai
Abstract
Finding a global optimal solution to the distribution network reconfiguration (DNR) problem in a short time is a challenging task. This study proposes a real‐time online data‐driven DNR (3DNR) method. Power loss minimisation, lowest bus voltage maximisation and reliability maximisation are taken as objectives. First, in this study, a methodology combining heuristic algorithm and metaheuristic algorithm to solve DNR is proposed. Then a set of data that satisfies the data drive model requirements is obtained. Next, the improved convolution neural network is used to train the data set of DNR. Unlike the state‐of‐art methods, the proposed 3DNR can realise the real‐time online reconfiguration without power flow calculation. The feasibility and effectiveness of the proposed method are demonstrated on IEEE‐34, IEEE‐123, and a practical distribution system in Taiwan.