Enhancing the resiliency of transmission lines using extreme gradient boosting against faults

Abstract

The study presented in the paper aims to reduce the relaying time associated with detection and classification of fault along with accurate fault location estimation with the help of Teager Kaiser energy operator (TKEO) and extreme gradient boost (XGBoost) algorithm. The complete analysis has been conducted on IEEE 14 Bus transmission network. Four XGBoost modules corresponding to each phase have been developed for fault detection, classification, and similarly four regression modules are developed for fault location estimation depending on the type of fault. The proposed methodology is able to precisely determine the fault, locate it and also recognize the faulty phase(s). Special attention has been paid in determining the high impedance fault (HIF) apart from usual low impedance faults (LIF) while varying the fault parameters such as evolving nature of faults, fault inception angle, power flow angle and change in transmission line parameters etc. Voltage and current measurements recorded at the relay location are utilized. Moreover, the proposed XGBoost algorithm-based approach is able to detect and classify faults in around 1–2 ms time. The absolute error in fault location is around 0–2 km in most of the scenarios. A comparative study clearly states the merits of the proposed scheme.