Detection and Characterization of Multiple Discontinuities in Cables with Time-Domain Reflectometry and Convolutional...

Area of Science:

• Electrical Engineering
• Signal Processing
• Machine Learning

Background:

• Cable integrity is crucial for reliable signal transmission.
• Accurate detection of impedance discontinuities is essential for fault diagnosis and maintenance.
• Traditional methods may lack precision in characterizing complex faults.

Purpose of the Study:

• To develop a convolutional neural network (CNN) for detecting and characterizing impedance discontinuity points in cables.
• To leverage time-domain reflectometry (TDR) signals for automated cable fault analysis.
• To quantify the type, position, and magnitude of impedance discontinuities.

Main Methods:

• A CNN model was designed to analyze TDR signals.
• The CNN was trained using numerous simulated signals generated by a calibrated transmission line simulator.
• The transmission line model was calibrated using stepped-frequency waveform reflectometry measurements.
• The trained CNN was validated on both simulated and real-world measured signals.

Main Results:

• The CNN successfully detected and characterized impedance discontinuity points.
• In experimental tests with capacitive faults, 100% detection accuracy was achieved.
• Position and magnitude of discontinuities were estimated with a root-mean-squared error of 13 cm and 14 pF, respectively.

Conclusions:

• The proposed CNN-based method offers a highly accurate and robust solution for cable impedance discontinuity detection and characterization.
• This approach enhances the reliability of cable fault diagnosis through automated signal analysis.
• The method demonstrates significant potential for practical applications in cable maintenance and monitoring.