Frequency-Domain Localization Method for High-Resistance Faults in Railroad Signal Cables Based on Active Noise Reduction

Fault detection of railroad signal cable is of great significance to the safety of railroad system. We propose an improved frequency-domain reflection (FDR) method based on simultaneous sampling of signal and noise for locating high-resistance faults, which is the most common fault in railroad signal cables. First, the return signal of the intact cable core and the inductive signal of the twisted pair are obtained by two-channel simultaneous sampling, and the two signals are Fourier transformed into spatial-domain waveforms. Second, the spatial-domain waveforms of the two signals have ambient noise and noise generated by multiple reflections at the transitions. We calculate the proportionality between the two spatial-domain waveforms by the least square method (LSM), denoted as the transmission factor. Finally, the fault line and its twisted pair are sampled synchronously during subsequent cable fault measurements. The fault line signal is subtracted from the product of the transmission factor and the twisted pair signal. The noise generated by the ambient noise and multiple reflections at the transfer is actively eliminated, highlighting the location of the fault and improving the localization effect of the original FDR method. A hardware system was designed based on this algorithm. The hardware system can transmit a 100-kHz–5-MHz frequency sweep signal to the cable, and the return signal required by the active noise reduction algorithm is collected by two-channel synchronized sampling and intermediate frequency (IF) sampling, and the data processing is subsequently carried out in the host computer. The improved algorithm proposed in this article achieves fault location for high impedance faults in cables below 1000 m, with a maximum resistance of 0.5 M $\Omega $ . However, the traditional FDR methods cannot obtain effective positioning results, which verifies the effectiveness and superiority of the proposed method.