A New ROP Prediction Method Based on CNN-LSTM-Attention

Rate of penetration (ROP) is a key factor in reducing drilling costs and optimizing drilling. However, the majority of ROP prediction methods are based on simple intelligent algorithms and traditional physical models, which ignore the relationship between deep sequences of drilling data, leading to lower prediction accuracy. To enhance the prediction accuracy of ROP, a ROP prediction model based on CNN-LSTM-Attention is proposed. Firstly, Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) are adopted to extract the features of deep sequence data. Then, the attention mechanism is introduced to effectively extract key information of sequence features. Finally, the model is constructed and verified on a publicly available dataset. The results indicate that the proposed model can reduce the root mean square error (RMSE) by 38.22 % and improve the Determination coefficient (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$R2$</tex>) by 21.51 % compared to the existing model.