In this proposal, we propose an FMCW radar based cable remaining useful life (RUL) estimator to detect, locate, and classify cable faults, and estimate the cable RUL. The system includes three key elements - an FMCW radar, a cable degradation model, and a machine learning (ML) based data processing module. The FMCW radar transmits a chirp down to the cable through a lead cable and an impedance matching network, receives the reflected signal from the faulty cable, converts it to the intermediate frequency (IF) band, filters the IF signal to compress the noise, converts the IF signal to the baseband, and collects baseband samples. The samples are then filtered with a filter bank and converted to spectrum domain to feed the ML algorithm for fault detection, property estimation and cable RUL estimation with the help of the cable degradation model. The proposed FMCW radar has the capability to minimize the impact of transmitter (Tx) leakage and cable noise to obtain a high detection sensitivity for soft fault sensing. Various grades cable faults are modeled in the cable degradation model for the ML model training for faults detection, classification, and RUL estimation. The advantages of the proposed FMCW radar based RUL estimator include (1) high sensitivity, (2) great noise impact reduction, (3) high accuracy of fault detection and RUL estimation with the cable degradation model and ML algorithm.
