Distributed photovoltaic (PV) hosting capacity is commonly used to assess the PV capacity allowed to be accommodated under the premise that the distribution network satisfies safe and stable operation. However

the existing hosting capacity calculation methods mostly rely on accurate network topology and line parameter information and have the limitations of large scene simulation and long time-consuming

which are difficult to be applied to low voltage distribution network (LVDN). Therefore

this paper proposes a data-driven method for distributed PV hosting capacity in LVDN. Firstly

an explicit neural network is applied to fit the mapping relationship between the power of each consumer and the voltage of nodes in LVDN and construct explicit equations reflecting the connection between the power and the voltage of the nodes according to the inference logic of the network. Secondly

calculate the sensitivity of each phase voltage to the node power and construct partition assignment index to realize adaptive partition under non-all-phase user access in LVDN. Finally

based on the clustering performance index and power flow fitting model

a hosting capacity calculation model is constructed for the LVDN considering voltage deviation and three-phase unbalance

enabling fast and efficient calculation of PV hosting capacity. The IEEE-123 node model and an actual LVDN is verified and the results show that the proposed method is not only applicable to the actual LVDN scenario but also has excellent robustness

which can approach the theoretical hosting capacity of the LVDN under the condition of incomplete topology and line parameter information.