Feature Extraction and Identification of Pipeline Intrusion Based on Phase-Sensitive Optical Time Domain Reflectometer

Since fiber distributed vibration sensing (DVS) system based on phase-sensitive optical time domain reflectometer (Φ-OTDR) has the characteristics of identifying intrusion signals, wide monitoring range and high system sensitivity, correct identification of intrusion types by the system is an important issue to promote the engineering of this technology. In this paper, based on the intrusion signal of Φ-OTDR system, a multi-dimensional feature extraction and selection method is proposed. The polynomial least squares method is used to remove the trend term from the vibration signal, and the wavelet threshold denoising method is used to reduce the noise interference. The short-time analysis in the time domain and the wavelet analysis in the wavelet domain are combined to extract the multi-dimensional characteristics of the signal. The feature selection is based on the QUICKREDUCT algorithm. The experimental results show that the feature vector obtained by this method is relatively complete, and it is less affected by the environment, and the recognition rate is higher, reaching over 92%.