A data-driven framework for HVDC fault detection, classification, and location using time domain decomposition and machine learning

High-voltage direct current (HVDC) transmission systems are an essential technology for long-distance bulk power transmission, and their reliable protection is crucial. A novel application of the Time Domain Decomposition (TDD) method is presented in this paper for the analysis of electrical transients in HVDC systems. Extracted characteristic “electrical mode shapes” from synchronized terminal measurements serve as a unique, sign-consistent spatial fingerprint for fault identification. Fault detection is achieved using an effective Z-score-based adaptive thresholding method applied to modal energy. Based on the mode shapes, feature vectors are extracted to train various machine learning classifiers, along with the sign-consistency enforcement method. Fault type classification is performed using a Support Vector Machine (SVM). At the same time, a regression tree model is employed to locate faults across a wide range of simulated fault types, fault resistances, and fault locations in a model HVDC system. The simulation results for the two-terminal VSC HVDC scheme show that the proposed TDD-based method provides a common platform for fault detection, classification, and location identification. The proposed method achieves 94.1% classification accuracy, a fault location MAE of 0.50 km, and a fault resistance estimation MAE of 2.6 Ω. A 5-fold Cross-Validation (CV) process conducted over an expanded dataset demonstrates the reproducibility of the proposed approach.

Authors

Institutions

Publication Details

Journal
Electric Power Systems Research
Published
2026-11-16
DOI
https://doi.org/10.1016/j.epsr.2026.114513
Primary Topic
HVDC Systems and Fault Protection
Type
article
Field-Weighted Citation Impact
0.00

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

A data-driven framework for HVDC fault detection, classification, and location using time domain decomposition and machine learning

K I-K Wang, Nirmal‐Kumar C. Nair, Abhisek Ukil, Leyla Zafari
1 links
Electric Power Systems Research
HVDC Systems and Fault Protection
article

A data-driven framework for HVDC fault detection, classification, and location using time domain decomposition and machine learning

K I-K Wang, Nirmal‐Kumar C. Nair, Abhisek Ukil, Leyla Zafari
article en

Abstract

High-voltage direct current (HVDC) transmission systems are an essential technology for long-distance bulk power transmission, and their reliable protection is crucial. A novel application of the Time Domain Decomposition (TDD) method is presented in this paper for the analysis of electrical transients in HVDC systems. Extracted characteristic “electrical mode shapes” from synchronized terminal measurements serve as a unique, sign-consistent spatial fingerprint for fault identification. Fault detection is achieved using an effective Z-score-based adaptive thresholding method applied to modal energy. Based on the mode shapes, feature vectors are extracted to train various machine learning classifiers, along with the sign-consistency enforcement method. Fault type classification is performed using a Support Vector Machine (SVM). At the same time, a regression tree model is employed to locate faults across a wide range of simulated fault types, fault resistances, and fault locations in a model HVDC system. The simulation results for the two-terminal VSC HVDC scheme show that the proposed TDD-based method provides a common platform for fault detection, classification, and location identification. The proposed method achieves 94.1% classification accuracy, a fault location MAE of 0.50 km, and a fault resistance estimation MAE of 2.6 Ω. A 5-fold Cross-Validation (CV) process conducted over an expanded dataset demonstrates the reproducibility of the proposed approach.

Electric Power Systems ResearchVol. 265
University of Auckland (NZ)
Ministry of Business, Innovation and Employment
Industry, innovation and infrastructure, Affordable and clean energy
HVDC Systems and Fault Protection
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon