Pantograph-Catenary Marginal Index Method Using ICEEMDANSPWVD for Railway Hard Spot Diagnosis

This study proposes a novel method for diagnosing hard spots in railway overhead contact lines using the Improved Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (ICEEMDAN) combined with the Smoothed Pseudo Wigner-Ville Distribution (SPWVD). Hard spots, caused by localized mass concentration, sharp bends, or improper equipment installation, significantly degrade the current collection quality in railway traction systems. Traditional methods for hard spot detection often struggle with non-stationary and nonlinear characteristics of pantograph-catenary vibration signals, as well as cross-term interference in time-frequency analysis. To address these challenges, this study introduces a hybrid framework that integrates ICEEMDAN for adaptive signal decomposition and SPWVD for time-frequency feature extraction. The proposed method establishes a three-phase collaborative mechanism: modal decomposition, time-frequency reconstruction, and marginal spectrum feature extraction. This approach enables precise separation of hard spot impact features and quantifies their energy characteristics, providing a robust tool for condition monitoring of pantograph-catenary systems. Through case studies involving measured pantograph vertical acceleration data, the method effectively identifies hard spots by analyzing the time-frequency distribution and calculating the Pantograph-Catenary Marginal Index (PCMI). Field inspections confirmed the presence of contact wire deformations at locations where the PCMI exceeded predefined thresholds, validating the diagnostic accuracy of the proposed method. Additionally, the stability and repeatability of the PCMI were verified across multiple runs at different speeds, demonstrating its practical value for guiding targeted railway maintenance and repair operations.