A Novel Chirp Detector Algorithm for Universal Pantograph Arc Detection

Résumé

Catenary-pantograph arc faults pose a serious threat to the reliability and safety of electric railways. Faults cause violent transient disturbances, accelerate equipment aging, and disrupt power continuity. Real-time applicability is limited for traditional approaches based on their reliance on image processing or vision-based deep learning and their computation latency is greater than the arc time constant. The paper introduces a novel image-free arc detection algorithm that directly processes measured pantograph current signals without relying on vision data. The novelty of the chirp-inspired algorithm lies in integrating band-pass filtering, differentiation, Hilbert transform envelope extraction, and multi-stage statistical processing to construct an efficient arc detection index. In contrast to existing approaches, the algorithm exploits the inherent physical fingerprints of arc transients in the electrical current waveform, representing the first systematic investigation into current-based arc detection in pantograph–catenary systems. The approach is evaluated across 20 test cases under varying arc time constants and fixed voltages to demonstrate its universality. The algorithm consistently distinguishes arc events from mechanical or noise-induced oscillations, providing a stable baseline after arc extinction and enabling real-time and reliable monitoring. The proposed image-free system surpasses vision-based schemes for arc detection, providing a scalable and feasible solution to leading-edge railway electrification systems.

Citer ce document

Accès au document

Texte intégral en lecture en ligne, réservé aux abonnés SPHAERO et aux membres de l'institution. Se connecter

Auteur(s)

Statistiques

Consultations : 1

Téléchargements : 0