Distribution network state estimation is based on distribution network (DN) topology identification (TI). The connection of high-penetration renewable energy, on the other hand, makes TI of DN more difficult. This manuscript therefore proposes an active distribution network (ADN) TI method based on a one-dimensional convolutional neural network. The characteristics of the nodes are analyzed in light of the sensitivity of node voltage to DN topology changes in order to select the key nodes where the distribution network phasor measurement unit (DPMU) should be placed. This can save money on investment and make model training less redundant. Using photovoltaic (PV) units and a modified IEEE-33 bus DN, a number of tests are carried out. Under limited DPMU measurement, the results demonstrate that the proposed distribution network topology identification method can achieve high accuracy TI in ADN.
HTML PDFShare this article
Physical Mathematics received 686 citations as per Google Scholar report
