Abstract: The automatic identification of feeder-consumer connectivity in low-voltage distribution network（LVDN） is the basis of intelligent LVDN. By analyzing the space, time and spatiotemporal characteristics of the voltage for load nodes in LVDN, a datadriven based identification method of feeder-consumer connectivity in LVDN is proposed, relying on the massive operation data of power grid produced by the development of the"Internet + "smart energy. An optimization model based on Kirchhoff’s current law is established to identify feeder-consumer connectivity, which is transformed into a quadratic programming problem. At the same time, in order to improve the solution efficiency and solve the identification problem for vacant users, an electricity meter clustering method based on voltage spatiotemporal characteristics is proposed. The effectiveness of the proposed method is verified on a case system with 73 users and multiple pilot LVDNs. Furthermore, the influence of meter measurement error and data length on the recognition results is investigated.