A phase space and network-based approach for diagnosing compensation capacitor faults in Jointless Track Circuits

Abstract

The reliability of compensation capacitors in Jointless Track Circuits (JTCs) is critical for maintaining stable signal transmission and ensuring train operation safety. However, these components are prone to degradation due to long-term use and environmental disturbances, leading to potential signal anomalies. Traditional diagnostic methods often fail to cope with the nonlinear, time-varying, and multi-fault characteristics of track circuit signals. This paper proposes a novel fault diagnosis approach that integrates phase space reconstruction with complex network analysis. First, one-dimensional track signals are mapped into a high-dimensional phase space to reveal dynamic behaviors. Singular Value Decomposition (SVD) is applied to the trajectory matrix for dominant feature extraction. A complex network is then constructed from the processed signal, and key topological metrics — such as degree centrality, clustering coefficient, betweenness centrality, and local structure entropy — are computed to identify fault-related patterns. Experimental validation using real-world JTC signal data demonstrates that the proposed method achieves superior diagnostic accuracy and robustness compared to conventional techniques. Notably, the clustering coefficient proves highly sensitive in differentiating between healthy and faulty conditions. The proposed framework offers a scalable and effective solution for early fault detection and real-time condition monitoring in railway signal systems.