Active incremental Support Vector Machine for oil and gas pipeline defects prediction system using long range...

Area of Science:

• Materials Science
• Mechanical Engineering
• Data Science

Background:

• Traditional "smart pig" systems for oil and gas pipeline inspection offer limited continuous monitoring and predictive capabilities for defects.
• The need for real-time, continuous pipeline condition monitoring is critical for preventing failures and ensuring operational safety.

Purpose of the Study:

• To propose and evaluate a novel technique for real-time pipeline defect prediction and condition monitoring.
• To address the limitations of existing inspection methods by enabling continuous surveillance.

Main Methods:

• Utilized long-range ultrasonic transducers (LRUT) for data acquisition.
• Implemented an active incremental Support Vector Machine (SVM) classification approach for real-time defect analysis.
• Collected 56 feature data points from a lab-scale experimental rig simulating pipeline defects.

Main Results:

• Achieved classification accuracy comparable to traditional batch training methods.
• Significantly decreased computational time for defect classification.
• Demonstrated the feasibility of continuous monitoring using LRUT and active incremental SVM.

Conclusions:

• The proposed LRUT and active incremental SVM technique offers an efficient and effective solution for real-time pipeline defect prediction.
• This method enhances pipeline condition monitoring by providing continuous surveillance and faster analysis.
• The approach shows promise for improving the safety and reliability of oil and gas infrastructure.