On the Fault Detection Technology of the Cable Based on the Empirical Mode Decomposition Filter

摘要

Once malfunctions of power cables occur, rapid and accurate determination of the fault location not only reduces the cost of repairing and the outage loss, but also improves the reliability of power supply. The detection signal of cable fault is susceptible to the noise of test field, so it is of difficulty to identify the characteristic points of the reflected pulse from the fault location, thus fails to meet the requirements of fault location. This paper makes an analysis of the law of the time-frequency distribution of the disturbed pulse signal with the Empirical Mode Decomposition (EMD) method, and also realizes how to determine the cut-off point of the components dominated by the noise and the useful signal in the component of eigenmode with the method based on autocorrelation function. Since this method is able to eliminate the useful signals in the dominant components in the noise, which has negative effect on the accuracy of the reconstructed signal, a method of wavelet filter is proposed to extract the useful signal from the component of eigenmode dominated by the noise. Finally, the results from numerical simulation validate this method in improving the performance of filtering.