Step Frequency TR-MUSIC for Soft Fault Detection and Location in Coaxial Cable

Soft faults in cables may trigger short circuits and open circuits in time, in that they ought to be detected and thus eliminated at the earliest possible stage, as to ensure safe and stable operation of the cables. A method called the time-reversal multiple signal classification (TR-MUSIC) had been proposed in the literature, which has been demonstrated to be an effective technique for locating soft faults in cables, due to its high resolution and excellent noise robustness. However, traditional TR-MUSIC relies on a vector network analyzer (VNA) for measuring the scattering matrix of cables, which adds cost and complexity to its implementation. In this regard, a new way of acquiring the desired scattering parameters is here proposed. An arbitrary function generator is used to inject incident signals into the cable under test (CUT), and an oscilloscope is used to collect the reflected signals. After postprocessing, the phase of scattering parameters can be obtained. There is another key issue in the image of the detection results, and ghost traces caused by the periodicity of Green’s function severely impact the vision saliency of the actual fault location, which limits the performance of the fault location. A step frequency variant of TR-MUSIC has been proposed, therefore, to mitigate ghost traces. Experimental results show that the fault location error of the proposed approach is smaller than 0.33% for a 51-m long coaxial cable. Moreover, in the case of introducing noise, the proposed approach can operate in situations with signal-to-noise ratios (SNRs) as low as −6 dB.