Aiming at the problem that it is difficult to effectively extract fault features and accurately identify faults when wind turbines fault, In this paper, a wind turbine fault diagnosis method based on improved SE-CNN is proposed. Firstly, a sliding window is used to expand the historical operating data of the wind turbine in fault collected by the supervisory control and data acquisition (SCADA) system, secondly, the improved squeezed excitation network (SEnet) is used to adjust the weights of the sample data, the global maximum pooling layer is used to improve the convolutional neural network (CNN), and finally the improved CNN is used to learn the fault features and perform fault diagnosis. The experimental results show that the improved SE-CNN outperforms RNN, PCA-DNN, and BiLSTM methods in fault diagnosis, which verifies the effectiveness of the proposed method in wind turbine fault diagnosis.