Bayesian Optimized Ensemble Decision Tree models for MT-VSC-HVDC Transmission Line Protection

Abstract

Over the last few decades, the High Voltage Direct Current (HVDC) technology has experienced significant growth. HVDC grid technologies are increasingly being employed for strengthening transmission systems and improving connectivity. In cases of long-range and bulk power transmission, HVDC systems have proven to be an attractive option compared to HVAC systems. HVDC grids exhibit reduced power loss and almost negligible lines reactive power. Faults must be fixed promptly, regardless of any challenges. This study presents a fault detection and classification method based on Bayesian optimized decision tree classifiers for an MT-VSC-HVDC transmission system. The primary objective of this research is to extract the DC voltage and current signal through the relays installed in the HVDC network. Afterward, fourteen features are formulated using these signals for the experimentation. Based on these features, Bayesian-optimized decision tree classifier is used to identify and differentiate the faults events. The proposed approach enables rapid identification, faster detection, and fixation of both internal and external faults. The proposed approach is rigorously assessed for various probable fault circumstances simulated with varying transmission system operating parameters. This experimental approach considerably reduces the complexity and time required to identify faults at various points on the HVDC transmission grids with high precision.