WITHDRAWN: Formation drillability prediction by using cascade model based on well logging data in the deep drilling process

Abstract In the process of deep drilling, obtaining the drillability information of the borehole is the primary premise to realize drilling planning and drilling control to achieve safe and efficient drilling. Therefore, a method of predicting formation drillability by well logging data based on the cascade model is proposed. First, the first-level model takes logging parameters as input, and the algorithm is mainly composed of multiple integrated learning algorithms and regression algorithms. Then, three types of model outputs with the best performance are selected as input to the second-level model. Stacking is used to form the second-level model, and the output is formation drillability. Numerical simulation results show that the cascade model with higher accuracy and generalization ability performs better than the regression algorithms such as XGB, KNN, SVR, GBDT, RF, etc., and can be used as a prediction model for drillability data-driven methods. This study provides a new way to study formation drillability in the era of artificial intelligence.