A Novel Hybrid Model for Online Prediction of Rate of Penetration (ROP) in Drilling Process

In this paper, a novel hybrid model is proposed for online prediction of rate of penetration (ROP) in drilling process, which including two parts (online data pre-processing and online hybrid modeling). In the first part, threshold filtering and Savitzky Golay (SG) filtering are both employed to enhance the quality of drilling data considering the expert experience and data characteristics. In the next part, a novel hybrid model with error compensation is established, which is combined the Bingham sub-model and gradient boosting decision tree (GBDT) sub-model. To better capture the dynamic changes of ROP, the hybrid model is updated with moving window strategy. Finally, compared simulation results with well-known ROP prediction models indicate the efficiency of the hybrid model.