Pattern Recognition Method For Heterodyne ?-OTDR Based On Down Conversion And IQ Demodulation

This paper proposed a high SNR,excellent real-time performance and high recognition rate ?-OTDR early-warning system,which could be used in long distance pipeline leakage monitoring.On the depth study of ?-OTDR system's development at home and abroad,this design used heterodyne receiving mode,analog down conversion,digital IQ demodulation and bp-neural network to implement the full embedded and real-time monitoring system.In heterodyne receive mode,the AOM introduced 200 M frequency shift.This design used analog down conversion circuit to remove the frequency shift,which also reduced the requirement of sampling rate.In traditional architecture,GHz DAQ and signal generator were used to generate modulation pulse and synchronize high speed data acquisition,which were improved to FPGA and peripheral circuit in this design to cut down system's cost and size.About signal processing,FPGA and DSP dual-core full embedded architecture was used to enhance real-time performance.In FPGA,digital IQ demodulation algorithm was used to get signal's amplitude.In DSP,moving average and difference was used to locate vibration signals,and image processing,morphological feature extraction,bp-neural network were used to classify vibration pattern.Compared to traditional mode recognition based on one dimension's features,this design's false positive rate was lower and the recognition rate was higher.This design's GUI,made by QT,could display location curve in real time and change system parameters quickly.By experiments,hardware circuit,software in FPGA and DSP could achieve intended purpose,and this design could monitor in real time with over 90% recognition rate.