Localization of Faults in Coaxial Cables using Time- Domain Reflectometry and Support Vector Machine

Reliable co-axial cable systems are crucial for safely operating electrical and electronic equipment. Detecting and localizing faults in these networks are of utmost importance to ensure proper functioning and prompt remedial actions. This paper explores the application of Time-Domain Reflectometry (TDR) in conjunction with a machine learning algorithm-based classification method to identify and localize faults in co-axial cable systems accurately. The proposed approach involves the generation of a narrow pulse using a signal generator and injecting it into the faulty cable, followed by capturing the reflected signal with an oscilloscope. The collected data is then used to train the machine learning algorithm for the purpose of fault identification. A total of 150 datasets were recorded; out of this, 120 were used for training purposes, and 30 were used for testing purposes. Cubic support vector machine was chosen as the machine learning algorithm, and a classification accuracy of 96.7 % was achieved.