FEATURE EXTRACTION OF TREE-RELATED HIGH IMPEDANCE FAULTS AS A SOURCE OF ELECTROMAGNETIC INTERFERENCE AROUND MEDIUM VOLTAGE POWER LINES' CORRIDORS

One of the faults in medium voltage (MV) overhead power line is a high impedance fault (HIF) owing to low-current discharge to a tree (THIF). This type of fault generates signals in wide frequency bandwidth which may lead to electromagnetic interference (EMI) with neighboring devices and consequently results in degradation in the performance of nearby systems. This problem becomes more critical when MV power lines path is located in a wooded area in which there will be frequent transient conflicts between trees and power lines especially in the windy conditions. Given the importance of this issue, the ability of THIF to generate EMI is first demonstrated in this paper. Thereafter, a hybrid technique based on combination of quantile regression (QR) and empirical mode decomposition (EMD) is proposed to perform a feature extraction from THIF signals. By comparing the QR results of different samples of THIF signal with other similar signals, the validation of proposed method is depicted. In summary, the original contributions of current research include 1) assessing EMI due to THIFs, 2) using EMD in pre-processing of THIFs signals and extracting their main components, 3) recommending QR for the feature definition of THIF.