Abstract:Intelligent drilling represents a future trend in petroleum drilling. Intelligent prediction and optimization of drilling rate of penetration (ROP) constitute pivotal research areas within intelligent drilling, promising enhanced drilling efficiency and cost reduction. The establishment of intelligent prediction models for ROP relies heavily on a vast amount of drilling data. However, data collection devices in oil fields exhibit instability, accompanied by significant environmental noise. Consequently, drilling data often contain numerous outliers and noise, which disrupt the training of ROP models and degrade their accuracy. In this paper, a methodology for handling anomalous data is proposed based on oilfield drilling data. Firstly, isolated forest (IF) detection is employed to identify and remove outliers, followed by the application of K-nearest neighbor (KNN) interpolation algorithm to address missing values. Finally, locally weighted scatterplot smoothing (LOWESS) is utilized to smooth the data and eliminate noise. Results indicate that the drilling speed prediction model built upon the processed data exhibits higher predictive accuracy. The proposed anomalous data handling methodology contributes to the enhancement of intelligent ROP prediction models, demonstrating practical utility. With the advancement of technology in the petroleum industry, intelligent drilling, as an efficient and intelligent drilling technique, has gradually garnered widespread attention (Lig, et al. 2022). Within this trend, intelligent prediction and optimization of drilling rate of penetration (ROP) have emerged as significant research areas in the field of intelligent drilling. Accurate prediction and optimization of mechanical drilling speed can significantly enhance drilling efficiency and reduce drilling costs, thereby injecting new vitality into the development of the petroleum industry (Barbosa, et al. 2019). However, in reality, we face numerous challenges. The complex and dynamic nature of oilfield environments, coupled with the instability of data collection devices and the influence of environmental noise, pose significant challenges to the quality of drilling data (Geekiyanage, et al. 2021). In such circumstances, drilling data often contain various outliers and noise, which not only have the potential to misguide decisions made by drilling engineers but also interfere with the training of intelligent prediction models for mechanical drilling speed, thereby reducing the accuracy of the models.

Intelligent Lost Circulation Monitoring Method Based on Data Augmentation and Temporal Models

Intelligent Monitoring Model for Lost Circulation Based on Unsupervised Time Series Autoencoder

Early Lost Circulation Monitoring Using a Hybrid CNN-LSTM Model

Intelligent Lost Circulation Early Warning Method Based on HALSTM Network

Lost Circulation Prediction in South China Sea using Machine Learning and Big Data Technology

Prediction of Drilling Fluid Lost-Circulation Zone Based on Deep Learning

Prediction of Drilling Leakage Locations Based on Optimized Neural Networks and the Standard Random Forest Method

Interpretable Lost Circulation Analysis: Labeled, Identified, and Analyzed Lost Circulation in Drilling Operations

Lost circulation monitoring using bi-directional LSTM and data augmentation

Real-time Monitoring and Diagnosis System of Lost-circulation Risk

Prediction of Lost Circulation in Southwest Chinese Oil Fields Applying Improved WOA-BiLSTM

Research on lost circulation diagnosis model based on smote-tomek and stacking ensemble learning

ANN Model for Predicting Mud Loss Rate from Unconfined Compressive Strength and Drilling Data

A Case Study on the Classification of Lost Circulation Events During Drilling using Machine Learning Techniques on an Imbalanced Large Dataset

Early Monitoring Method of Downhole Accident Driven by Physics Based Model and Data Driven Methods Coupling

Automatic Implementation Algorithm of Pressure Relief Drilling Depth Based on an Innovative Monitoring-While-Drilling Method

Hybrid machine learning algorithms to enhance lost-circulation prediction and management in the Marun oil field

A realtime drilling risks monitoring method integrating wellbore hydraulics model and streaming-data-driven model parameter inversion algorithm

An Identification Method for Intelligent Prediction of Drilling Overflow Based on BiLSTM-BP Neural Network

Enhancing Intelligent Drilling Rate of Penetration PredictionModeling Through Anomalous Drilling Data Handling Methodology

A Real Time Monitoring Method for Well-Kick and Lost Circulation Based on Distributed Fiber Optic Temperature Measurement