Recurrent Graph Transformer Network for Multiple Fault Localization in Naval Shipboard Systems

TL;DR

This work tackles fault localization in complex naval MVDC 12kV shipboard power systems by introducing a temporal recurrent graph transformer network (RGTN) that combines GRU-based temporal modeling with graph-transformer-based spatial reasoning. The method processes time-series currents from 20 locations to classify faults across 10 possible labels, including multiple simultaneous faults, using cross-entropy training. Key contributions include integrating GRU with Graph Transformer Networks to capture both temporal and long-range spatial dependencies, validating on a four-zone MVDC shipboard model with line-to-line fault scenarios, and demonstrating superior localization accuracy (up to ~99.7%) and noise robustness compared with several baselines. The approach advances fault management in naval power systems by enabling precise, component-level fault detection and rapid, robust localization, with potential for HIL integration and enhanced naval energy resilience.

Abstract

The integration of power electronics building blocks in modern MVDC 12kV Naval ship systems enhances energy management and functionality but also introduces complex fault detection and control challenges. These challenges strain traditional fault diagnostic methods, making it difficult to detect and manage faults across multiple locations while maintaining system stability and performance. This paper proposes a temporal recurrent graph transformer network for fault diagnosis in naval MVDC 12kV shipboard systems. The deep graph neural network uses gated recurrent units to capture temporal features and a multi-head attention mechanism to extract spatial features, enhancing diagnostic accuracy. The approach effectively identifies and evaluates successive multiple faults with high precision. The method is implemented and validated on the MVDC 12kV shipboard system designed by the ESDRC team, incorporating all key components. Results show significant improvements in fault localization accuracy, with a 1-4% increase in performance metrics compared to other machine learning methods.

Figures (8)

• Figure 1: The multi-head attention in a Graph Transformer Network
• Figure 2: The Architecture of Recurrent Graph Transformer Network (RGTN) for Fault Localization in the MVDC shipboard system
• Figure 3: The 4-zonal MVDC 12kV system team2017model
• Figure 4: The ESRDC MVDC 12kV System Diagram with Measurement Locations and Fault Scenarios
• Figure 5: The R-GTN training accuracy for MVDC fault location