Abstract: With the acceleration of the intelligent process of coal mine, the accuracy and real-time performance of fault diagnosis of coal mine equipment are more and more critical, which plays an important role in ensuring production efficiency and the safety of miners. However, in the actual fault diagnosis work, there may be a variety of problems such as data missing, data anomaly and network transmission, which affect the real-time and high accuracy performance of fault diagnosis. Therefore, in order to meet the requirements of heterogeneous fault signals of coal mine equipment and obtain real-time state feedback, an intelligent fault diagnosis method for coal mine equipment based on cloud-edge collaboration is proposed. This method combines the more powerful computing capacity of the cloud server and the close transmission distance of the edge server to provide a more efficient and accurate fault diagnosis solution. This method preprocesses the monitoring data of coal mine equipment at the edge end. With the help of random forest algorithm, the isolation forest algorithm, transform domain filtering method and other algorithms, the noise in the data is removed, the missing values are filled, and the outliers are identified to ensure the integrity and accuracy of the data. After these preprocessing steps, the data is sent to the cloud server. After receiving the data, the cloud server further trains the universal fault diagnosis model. Firstly, the local features of the data are automatically extracted by the convolutional neural networks (CNN). Then, the Bi-directional gated recurrent unit (BiGRU) is used to learn the temporal dependencies of the data. The convolutional block attention module (CBAM) is combined to assign weights to different features to improve the diagnostic performance of the model. After the cloud server completes the training of the complex model, the trained model parameters are sent to the edge server in time. At the edge server, the lightweight learning model ShuffleNet V2 updates its parameters according to those received from the cloud server. After the update, it can quickly adapt to the patterns learned by the cloud-trained model and perform real-time fault diagnosis on coal mine equipment. Through simulation experiments, the effectiveness of the proposed method is proved, and the accuracy of real-time fault diagnosis at the edge server can be improved to 99.7%. This high accuracy indicates that the proposed cloud-edge collaborative intelligent fault diagnosis method can timely identify different types of faults in coal mine equipment, which can effectively ensure the safety and efficient operation of coal mine production.