Sticking is a serious accident for drilling, with various types and causes. If not detected and prevented in a timely manner, it will require a lot of time and processing costs, and even lead to wellbore scrapping. By predicting the probability of drill pipe sticking, targeted measures can be taken in advance to prevent and reduce drill pipe sticking incidents, ensuring the safety and continuity of drilling operations. An intelligent prediction method for stuck drilling based on the fusion of physical and data dual drivers has been proposed. This method adopts a series fusion mode, based on known physical laws and equations, to use the output of the physical model as a partial input of the data-driven model, in order to improve the physical interpretability and diagnostic accuracy of the data-driven model. In order to achieve the prediction of stuck drill accidents, a time series model is combined with a fused intelligent prediction model to predict future related parameter data and make stuck judgments; At the same time, an incremental learning model is introduced to preserve historical data and features, and incremental adjustments are made in combination with new data to achieve real-time updates and predictions of the model, achieving true real-time prediction. After the establishment of an intelligent diagnosis model driven by a combination of physics and data, accuracy, precision, and recall are used to evaluate the model. The field data of a well is used for the test, and the results show that the evaluation indexes of the intelligent diagnosis model of sticking based on the hybrid drive of physics and data proposed in this study are improved compared with the single model. The parameter data predicted by time series was substituted into the hybrid driven drill prediction neural network model for testing, and the predicted probability of drill sticking was consistent with the actual situation. After adopting the incremental learning model, the accuracy of the drill prediction model has been further improved. By combining time series prediction methods with intelligent prediction models under hybrid driving, the stuck diagnosis model has been upgraded to a stuck prediction model. The incremental learning model can update the model in constantly changing data streams, improve the accuracy and reliability of the model, and has good scalability and portability in actual drilling, which is of great significance and application value.