Stuck Pipe Detection in Oil and Gas Drilling Operations Using Deep Learning Autoencoder for Anomaly Diagnosis

Applied Sciences, Journal Year: 2025, Volume and Issue: 15(9), P. 5042 - 5042

Published: May 1, 2025

Stuck pipe events remain a critical challenge in oil and gas drilling operations, leading to increased non-productive time substantial financial losses. Traditional detection methods rely on manual monitoring expert judgment, which are prone delays human error. This study proposes deep learning autoencoder-based anomaly diagnosis approach enhance the of stuck incidents. Using high-resolution series data from Volve field, autoencoder model was trained exclusively normal conditions learn operational patterns detect deviations indicative events. The proposed leverages reconstruction error as an metric, effectively distinguishing between cases. results demonstrate that achieves accuracy 99.06%, with area under receiver operating characteristic curve (AUC) 0.958. Additionally, attained precision 97.12%, recall 91.34%, F1-score 94.15%, significantly reducing false positives negatives. findings highlight potential learning-based approaches improving real-time detection, offering scalable cost-effective solution for mitigating disruptions. research contributes advancing intelligent systems industry, risks, enhancing efficiency.

Language: Английский

Intelligent Thermal Condition Monitoring for Predictive Maintenance of Gas Turbines Using Machine Learning

Machines, Journal Year: 2025, Volume and Issue: 13(5), P. 401 - 401

Published: May 11, 2025

Gas turbines play a crucial role in power generation and aviation, where effective maintenance strategies are essential to ensure reliability. Traditional condition monitoring methods often rely on scheduled inspections, leading potential downtime increased costs. This study presents an AI-driven approach for thermal the predictive of gas using machine learning. An Extreme Gradient Boosting (XGBoost)-based classification model was developed distinguish between healthy faulty operating conditions based load data. The dataset, collected over six months from strategically placed thermocouples exhaust section, processed extract key statistical features such as mean temperature, standard deviation, skewness. proposed XGBoost achieved accuracy (CA) 97.2%, with F1-score 96.8%, precision 97.5%, recall 96.1%, demonstrating its effectiveness detecting anomalies. results indicate that integration learning turbine significantly enhances fault detection capabilities, enabling proactive reducing risk critical failures. provides valuable insights data-driven strategies, optimizing operational efficiency extending lifespan components. Future work will focus real-time deployment further validation extended datasets.

Language: Английский