The inter-provincial electricity spot market(IPESM) tariff mechanism is different from the calculation of tariffs under the traditional spot market, which is mainly affected by the declared price and declared capacity of market players, transaction paths, transmission capacity of the channel as well as the ratio of supply and demand, and the research on the prediction of transacted tariffs in the IPESM is of great significance. In this paper, an method based on a graph convolutional network is proposed for predicting transaction fees. Firstly, the input matrix of the prediction model is established by considering the inter-provincial trading paths and the information of market players. Secondly, the GCN was combined with the IPESM trading model for the prediction of electricity turnover and turnover price. Finally, the model prediction results were evaluated using MAPE and MAE. The simulation results show that the graphical neural network can effectively identify the mapping relationship between the characteristics of trading channels and declaration volume and the transacted electricity price in IPESM, and the results have high accuracy.