Abstract: Stick-slip vibration is an important limiting factor affecting drilling speed, safety and cost. The establishment of a reliable stick-slip vibration classification model is very important for oil drilling decision-making. A new method based on Bayesian optimization and eXtreme Gradient Boosting (XGBoost) is proposed to evaluate the severity of stick-slip vibration near the bit. The classification processing of the near-bit stick-slip vibration data is carried out. The main feature vectors of the original data is extracted through time domain and frequency domain analysis. A stick-slip vibration level identification and prediction model based on XGBoost is established, and Bayesian algorithm is uesd to the XGBoost hyperparameter tuning. Based on the model evaluation indicators of accuracy, recall, and F1 score, the test set is used to compare the optimized XGBoost model with the Classification and Regression Tree (CART) model and the Random Forests (RFs) model, respectively. The research results show that the method can effectively evaluate the stick-slip vibration level of the near-bit bit and has a high recognition accuracy.

Key words: stick-slip vibration, XGBoost, Bayesian optimization, risk assessment, near-bit