Signal post-processing method and device for improving meeting vehicle identification accuracy

Abstract

This invention discloses a signal post-processing method and apparatus for improving the accuracy of train passing identification, belonging to the field of high-speed railway condition monitoring technology. Based on candidate train passing events initially identified by existing methods, the method first performs time-domain waveform screening to eliminate events whose time-domain waveforms do not conform to a preset impact pattern. Then, for the screened events, the frequency domain features of their vibration signals are extracted. Finally, a second screening is performed based on the comparison results of the frequency domain features and a preset threshold to distinguish real train passing events. This invention, through a two-stage screening architecture of first the time domain and then the frequency domain, can effectively capture and utilize the essential differences in time-domain waveforms and frequency-domain energy distribution between transient train passing impacts and persistent interference, thereby significantly improving the accuracy and reliability of train passing identification and producing a high-purity train passing event dataset.

Images