Research on RBF neural network for pattern recognition and classification in Φ-OTDR systems

Phase sensitive optical time domain reflectometer (Φ-OTDR) has the advantages of higher sensitivity and spatial resolution, longer monitoring distance and multi-point positioning, and has been widely used in dynamic sensing fields such as perimeter security, rail transit and pipeline monitoring. Based on the Radial Basis Function (RBF) neural network, we extract, recognize, and classify the various constructed features of fiber optic vibration sensing signal including time domain, statistical domain, frequency domain. We accurately identify and classify the no intrusion events, shaking events, crossing events, and knocking events, with an average recognition rate of 93.85%. Our work points out a new direction for improving the recognition accuracy of Φ-OTDR systems for disturbance signals and decreasing false positive rate.