Distributed optical-fiber vibration sensor for the identification of pipeline leakage using relevant vector machine

A novel pattern recognition method based on relevant vector machine for optical fiber vibration sensor is proposed to meet the requirement of accurate identification of long-distance pipeline leakage. The sensing system is based on Sagnac interferometer, which is suitable for vibration signal detection. The initial vibration signals are firstly pretreated by the wavelet de-noising method. Then, the wavelet decomposition algorithm is utilized to obtain decomposition coefficients at all levels to build feature vectors. Furthermore, the relevant vector machine is presented as a classifier to identify three types of vibration signals, including pipeline leakage, human tapping and human walking. Experimental results demonstrate that the proposed method can discriminate three threat events with an average recognition rate above 85% with a 4-km-long sensing fiber, which can satisfy actual application requirements.