Event identification based on random forest classifier for Φ-OTDR fiber-optic distributed disturbance sensor

To reduce the nuisance alarm rate (NAR) for phase-sensitive optical time-domain reflectometer (Phi-OTDR) fiberoptic distributed disturbance sensors, an effective event identification method based on a random forest (RF) classifier is proposed in this paper. Through learning the features of time-domain disturbance signals using a random forest classifier, four kinds of disturbance events, including three kinds of real disturbance events, namely, watering, knocking and pressing, and one no-disturbance event, can be recognized effectively. The experimental identification rates for watering, knocking, pressing and no-disturbance events reach 93.79%, 97.36%, 97.06% and 98.12%, respectively. Experimental results indicate that this method based on a random forest classifier has high accuracy for event identification with an average identification rate of 96.58%.