A Fault Diagnosis Method for Oil Well Electrical Power Diagrams Based on Multidimensional Clustering Performance...

Area of Science:

• Petroleum Engineering
• Artificial Intelligence
• Signal Processing

Background:

• Traditional dynamometer cards for downhole monitoring in oilfield extraction have limitations in real-time performance and cost.
• Modern extraction demands more efficient and accurate condition monitoring solutions.

Purpose of the Study:

• To develop a novel fault detection method for beam-pumping units using motor power parameters.
• To enhance real-time monitoring capabilities and reduce maintenance costs in oilfield operations.

Main Methods:

• Dynamic mathematical modeling of pumping units to transform indicator diagrams into electric power diagrams.
• Extraction of fourteen new features from electrical parameters across time, frequency, and time-frequency domains.
• Development of a new effectiveness evaluation for Fuzzy C-Means (FCM) clustering integrating fuzzy membership and geometric structure.

Main Results:

• Analysis revealed distinct electric power diagram characteristics for five typical operating conditions.
• The proposed method achieved a fault diagnosis accuracy of 98.4%, outperforming SVM and ELM.
• The novel FCM evaluation function accurately determined optimal cluster numbers and improved clustering performance on UCI datasets.

Conclusions:

• The proposed motor power analysis method offers significant advantages in real-time performance, diagnostic accuracy, and cost-effectiveness for oilfield extraction.
• This approach effectively addresses the limitations of traditional monitoring methods, enhancing fault diagnosis capabilities.
• The method enables efficient real-time monitoring of beam-pumping unit wells.