Abstract
Abstract
In this paper, an efficient method is proposed for detecting and locating a soft fault, i.e. a physical degradation, in a Y-shaped network of unshielded twisted pair cables. The method uses jointly the observability analysis and time-frequency domain reflectometry experimental measurements. The fault detection consists in checking the consistency between the known characteristics of the cable under no-fault and faulty conditions, with the maximum value of a normalized time-frequency cross-correlation function computed on a sliding time window whose length is equal to the boundary exact observability time. The cable transmission speed and the wired network observability time are combined to locate precisely the fault. Soft faults are artificially created by connecting resistors of different values to the test cable. Experimental results show the efficiency of our fault detection and localization method.
Subject
Applied Mathematics,Instrumentation,Engineering (miscellaneous)