A Grid-Side Data-Driven Multigranularity Feature Fusion Model for Accurate Multitype Fault Diagnosis in Electric Vehicle DC Charging Piles
IEEE Transactions on Instrumentation and Measurement · 2026 · 75 · 3520814-3520814
DOI: 10.1109/tim.2026.3712926 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Instrumentation and Measurement, 2026. Verfasst von Li Liu und 5 weiteren Autor:innen. Die Arbeit nennt 29 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Natural Science Foundation of Guangxi Zhuang Autonomous Region.
Abstract (Originalsprache)
The fault of electric vehicle DC charging pile has attracted users’ attention to safety issues. Traditional fault diagnosis often requires front- and rear-stage fault features, and requires many characteristic parameters. Even though fault diagnosis methods are difficult to fully capture and identify multitype fault information. In order to reduce the demand for characteristic parameters and realize multitype fault diagnosis, this article proposes a new method, namely a grid-side data-driven multigranularity feature fusion (GSD-MGFF) model. This method only collects four characteristic information of the three-phase current and single-phase input line voltage of the charging module on the front-stage power grid side. It can reduce the complexity of the multitype fault diagnosis system and hardware sampling cost and facilitate the safety monitoring of the charging pile by the grid. The feature information is fused by the grid-side feature information fusion module. Then, the deep learning parallel space and global context feature extraction module is used to extract high-dimensional features. This module can fully capture and identify multitype fault information and avoid artificial feature selection errors. Finally, the fusion classification of spatial and global context features is realized by the spatial-context feature fusion classification module. To achieve the goal of accurate diagnosis of multitype fault front and rear stages of the DC charging pile. The simulation results show that this method has higher diagnostic accuracy, generalization ability, and robustness than the other comparison models. In the Typhoon hardware-in-the-loop (Typhoon HIL) experimental test, the diagnostic accuracy rate reached 96.46 %, under different fault scenarios.
Förderung
- National Natural Science Foundation of China
- Natural Science Foundation of Guangxi Zhuang Autonomous Region
Zitieren
APA
Liu, L., Li, Z., Zhao, R., Feng, Y., Wu, T. X., & Li, S. (2026). A Grid-Side Data-Driven Multigranularity Feature Fusion Model for Accurate Multitype Fault Diagnosis in Electric Vehicle DC Charging Piles. IEEE Transactions on Instrumentation and Measurement, 75, 3520814-3520814. https://doi.org/10.1109/tim.2026.3712926
Vancouver
Liu L, Li Z, Zhao R, Feng Y, Wu TX, Li S. A Grid-Side Data-Driven Multigranularity Feature Fusion Model for Accurate Multitype Fault Diagnosis in Electric Vehicle DC Charging Piles. IEEE Transactions on Instrumentation and Measurement. 2026;75:3520814-3520814. doi:10.1109/tim.2026.3712926
BibTeX
@article{Liu2026,
author = {Liu, Li and Li, Zhao and Zhao, Runqi and Feng, Yazheng and Wu, Thomas Xinzhang and Li, Shenwang},
title = {A Grid-Side Data-Driven Multigranularity Feature Fusion Model for Accurate Multitype Fault Diagnosis in Electric Vehicle DC Charging Piles},
journal = {IEEE Transactions on Instrumentation and Measurement},
year = {2026},
volume = {75},
pages = {3520814--3520814},
doi = {10.1109/tim.2026.3712926}
}
EUVIMED