Aiming at the limitation of traditional power system fault identification model, this paper proposed a transmission line fault identification algorithm based on two-dimensional Convolutional Neural Network (2D‐CNN), which can automatically extract fault characteristics of time‐domain signals of electric quantity and avoid errors caused by manual extraction of characteristic values. This paper taked IEEE 30 nodes power system as an example, collecting vast amounts of fault signal, the original deviation standardization will be one dimensional time domain signal by two‐dimensional gray scale, using two‐dimensional CNN forward training and reverse fine‐tuning, vector adaptively adjusted with Adam algorithms, introducing the mechanism of dropout, Softmax as top classifier, realization of transmission lines, including open circuit fault and open circuit fault, 17 kinds of fault classification and fault location. The experimental results show that the classification accuracy of the proposed method can reach 100%, which is higher than that of the traditional fault identification model, and can maintain a good classification effect when the noise level is increased. In addition, when the topological structure of power system is changed, this method can also achieve high recognition accuracy.