A method of recognizing faults in noise barriers of high‑speed railways was proposed based on the K nearest neighbors （KNN） algorithm and the phase‑sensitive optical time domain reflectometry （φ‑OTDR） system. A V‑shaped laying method of optic fiber cable was designed to sense vibrations of sound absorption boards at different heights of the noise barrier. And vibration signals under air turbulent force were acquired by the φ‑OTDR system. After the multi‑domain feature extraction and KNN classification of vibration signals， the state of noise barriers could be recognized. Results of the experiment showed that average recognition accuracy of 90.9% could be obtained even under complex field environments. This method could provide a feasible technical route for the fault detection of noise barriers， which could reduce dependence on professionals， so as to play an important role in improving the level of intelligent operation and maintenance.