The voltage prediction of the station area is the key to realize the timely management of the low voltage problem in distribution network. Most of the existing voltage prediction methods rely on the collection of power grid topology parameters and electricity information, which have the disadvantages of complicated required data, poor real?time performance, and large prediction errors. Therefore, this paper proposes a voltage prediction method based on the combined LSTM?BP model. Firstly, by analyzing the low?voltage formation mechanism in the station area, it is found that the dominant factor affecting the bus voltage is the users’ power consumption. Then the LSTM neural network is used to realize the short?term forecasting of load curves in the distribution station area, and the self?learning ability of BP neural network is used to establish the mapping relationship between users’ power and users’ voltage. By effectively combining the above two neural network models, the historical load data can be used to accurately and quickly predict the future low voltage situation in the station area. Finally, taking a station area as the research object and comparing the actual voltage data with the voltage data predicted in this paper, the results show that within the 1V error range, the prediction accuracy is 99.8%. Compared with the traditional voltage prediction method, during online implementation, the method proposed in this paper can realize real?time and accurate low voltage prediction without the need of information such as station topology, line parameters, users’ voltage, etc.