Speed-synchronized time-frequency ridge extraction from axlebox vibrations

Track irregularities critically impact railway operational safety, necessitating quantitative analysis. Existing geometric methods perform spatial domain analysis via angular domain resampling but rely heavily on precise synchronous speed data. Engineering challenges—such as sensor clock asynchrony, sampling discrepancies, and motion tracking delays—prevent reliable spatiotemporal synchronization between speed and vibration responses, causing phase shifts during feature separation. To address this without external sensors, we propose a method that synchronizes speed by extracting time-frequency ridge lines from axlebox vibrations. First, the Octave Linear Chirplet Transform calculates the time-frequency representation to obtain an initial ridge line. Then, based on the initial ridge line information, more accurate ridge line extraction parameters are obtained, and the time-frequency ridge line is further corrected by the Adaptive Fast Path Optimization algorithm. Finally, abnormal ridges are corrected using the ridge-speed proportional relationship and line conditions, enabling synchronous speed signal extraction directly from vibration data. The method’s efficacy is validated through simulations and experiments.