High Accuracy Intrusion Pattern Recognition using a Dual-Stage-Recognition Network for Fiber Optic Distributed Sensing System

Abstract

A dual-stage-recognition network combined with the fiber optic DAS system is designed to realize a high accuracy intrusion pattern recognition in practical complicated environments. Field test is performed to verify an average recognition accuracy rate of 97.04% for five typical events including animals, human and mechanical movement.