Research on Defect Localization of Power Cables Using Compressed Sensing-Based FDR Method

Frequency Domain Reflectometry (FDR) is widely used for detecting and locating local defects in power cables. However, noise, multiple reflections, and spectral leakage can generate spurious peaks and sidelobes in the FDR spectrum, which may mislead defect localization. Moreover, FDR testing requires the collection of a large volume of data, placing high demands on on-site storage and real-time processing. This study proposes a compressed sensing (CS)-based method for defect localization in power cables. The method first preprocesses the raw signal using Variational Mode Decomposition (VMD) and windowing and performs compressed sampling with a sparse random measurement matrix, after which the signal is reconstructed and preliminarily denoised under the Discrete Fourier Transform (DFT) basis, followed by sparse reconstruction under the Chirp Z-Transform (CZT) basis to obtain a sparse representation. Simulations and field tests verify that the proposed method maintains high localization accuracy even under high compression ratios, while effectively suppressing spurious peaks and sidelobe interference.