Predicting Bottom Hole Pressure in Controlled Pressure Drilling Using the BI-LSTM Classifier

Bottom-hole pressure is the amount of pressure at the hole's bottom, expressed as tons per square inch. The bottom hole level in a well that is flowing is equivalent to the amount of pressure drop in the tubing plus the well head level. Bottom hole level refers to the aquifer or deposit level at the bottom of the hole. The Bi-LSTM classification is proposed in this research to forecast bottom hole stress in monitored digging. The bottom-hole pressure (BHP) data is originally employed. The procedure then puts in the data quality evaluation components. Incomplete numbers, copies, anomalies, incompatible kinds of information, and data correctness are all checked. After the information's quality assessment, choosing features occurs. The method of limiting the input parameter to a model by utilizing only pertinent data and removing noise from the data is known as feature gathering. It is subjected to two-way evaluation and instruction. The greater amount of training information the algorithm has, the more accurate predictions it makes. Testing Data is utilized to assess the efficacy of the model. The algorithm enhances its forecasts by learning from the training data. Following that, the data is sent via bidirectional long-short-term memory (Bi-LSTM). In terms of memory, it outperforms typical recurrent neural networks. LSTMs do very well when they have a firm grasp on learning particular structures. Loss convergence occurs when the BiLSTM classification is followed. During instruction, it obtains a condition in which the loss settles to within the margin of error of the final result. Python software is used to carry out this work. In comparison graph the accuracy and voltage specificity of the proposed classifier as 92% and 92.4% is obtained.