The recent progress in the area of signal processing has led to the development of intelligent schemes for fault classification and faulty phase selection in distribution system. The conventional algorithms find limitations to detect high impedance fault (HIF) and HIF remains a great concern for utilities as it can cause serious danger and accident if not detected properly. This study presents an algorithm for the detection and classification of HIF in a multi-feeder radial distribution system based on the calculation of power spectral density of faulty current signals. Discrete wavelet transform has been used to decouple the time information from the frequency information to detect and classify HIF effectively. The proposed technique has been extensively assessed under various dynamic situations including the presence of distributed generation and nonlinear load. This technique does not involve any training or learning process nor it involves any concern about data synchronization. The fault detection time is improved as it requires only half cycle of post fault current signal to detect and classify faults successfully. It is noteworthy to mention that the proposed technique considers only one end terminal current data. The comparative assessment reveals that this method can overcome the limitations of other existing techniques.

信号处理领域的最新进展导致了配电系统故障分类和故障选相的智能方案的发展。常规算法发现了检测高阻抗故障（HIF）的局限性，HIF对于公用事业仍然是一个重大问题，因为如果未正确检测，它会导致严重的危险和事故。这项研究提出了一种基于故障电流信号功率谱密度计算的多馈线径向配电系统中HIF的检测和分类算法。离散小波变换已用于将时间信息与频率信息解耦，从而有效地检测和分类HIF。在各种动态情况下，包括分布式发电和非线性负载的存在下，对提出的技术进行了广泛的评估。该技术不涉及任何培训或学习过程，也不涉及数据同步。故障检测时间缩短了，因为仅需要故障后电流信号的半个周期即可成功检测和分类故障。值得一提的是，所提出的技术仅考虑一个终端电流数据。比较评估表明，该方法可以克服其他现有技术的局限性。故障检测时间缩短了，因为仅需要故障后电流信号的半个周期即可成功检测和分类故障。值得一提的是，所提出的技术仅考虑一个终端电流数据。比较评估表明，该方法可以克服其他现有技术的局限性。故障检测时间缩短了，因为仅需要故障后电流信号的半个周期即可成功检测和分类故障。值得一提的是，所提出的技术仅考虑一个终端电流数据。比较评估表明，该方法可以克服其他现有技术的局限性。