Distributed vibration sensors (DVS) based on phase-sensitive optical time-domain reflectometer (Φ-OTDR) has the advantage of long-distance real-time detection and can be used for railroad perimeter security monitoring. In order to meet the requirements of real-time railroad security monitoring, the vibration event signals occurring along the railroad line are extracted in time. In this paper, based on the weighting band-partitioning spectral entropy (WBSE) and short-time amplitude (STA) of the sensing signal detection, and the use of the improved spectral subtraction based on multi-taper spectral estimation for signal denoising, in order to realize the effective extraction and denoising of the vibration signal segment. In order to verify the effectiveness of the scheme, field tests of four common vibration events (train running, knocking fence, machinery vibration and pedestrian walking) at the railroad perimeter were conducted. The results show that the average detection accuracy of the four common vibration events reaches 88.39%, the signal-to-noise ratio (SNR) is improved to a range from 3dB to more than 9dB, and the overall processing time of the signal per 100 frames is about 0.24 s. These indexes indicate that the scheme can effectively meet the demand of real-time extraction of vibration event signals along the railroad.
Research on signal preprocessing method of railroad perimeter vibration events based on distributed vibration sensors
International Conference on Optoelectronic Materials and Devices (ICOMD 2024) ; 2024 ; Chongqing, China
Proc. SPIE ; 13549 ; 1354921
10.03.2025
Aufsatz (Konferenz)
Elektronische Ressource
Englisch