Key Technologies Research On Signal Detection And Recognition In ?-OTDR Distributed Optical Fiber Vibration Sensing System

As a new type of sensing technology,the phase-sensitive optical time domain reflection(?-OTDR)distributed optical fiber sensing technology has the advantages of good environmental tolerance,low energy consumption,high sensitivity,long monitoring distance compared with traditional sensing technology,and has been the emphasis and hotspot of the research.Among the manifold application fields of ?-OTDR distributed optical vibration sensing system,perimeter security is one of the most popular and most advantaged applications.The main task of perimeter security is to effectively detect and identify the vibration signals appearing in the system.But in long-distance monitoring,due to the complicated and changeable environment,it is difficult to effectively detect and identify external vibration signals.The requirement of execution efficiency further increases the difficulty of detection and identification in the long-distance monitoring.Therefore,this paper focuses on the efficient and accurate detection and identification of vibration signals in long-distance perimeter monitoring of ?-OTDR distributed optical fiber vibration sensing system.Firstly,the system principle of ?-OTDR distributed optical fiber vibration sensing system is introduced.The characteristics of fiber optic signals in this system and the effect of spectral subtraction noise reduction is analyzed and compared.For the long-distance perimeter monitoring of ?-OTDR distributed optical fiber vibration sensing system,a two-level detection algorithm based on double update parameters time-space detection and H-WELM is proposed.In the first-level detection,time and space domain detection algorithm utilizes double update parameters to maintain the validity of the detection relative to the single update parameter when continuous interference occurs,and reduces the false alarm on the spatial detection points in combination with the spatial domain detection.In the second-level detection algorithm based on H-WELM,once the signal is detected by the first-level algorithm,five characteristics of the current frame and the subsequent frame are extracted,and the weight of the H-WELM is controlled by the energy of vibration signals and false alarm signals.The result of the H-WELM is used for the final discrimination.By comparison,the second-level detection algorithm based on H-WELM has faster training and discriminating speed,and achieves better performance in distinguishing the vibration signal and the false alarm signal.The detection and false alarm rate are well balanced,as the joint effect of these two levels of detection algorithms.In view of the difficulty of recognizing the pattern of vibration signals efficiently and accurately,a vibration signals pattern recognition method is proposed based on multi-scale one-dimensional convolutional neural network(MS 1-D CNN).Firstly,the raw vibration signals are pre-processed,including pre-emphasis filtering,normalization,and spectral subtraction.The pre-processed signals are inputted to the inputs of MS 1-D CNN directly to realize the end-to-end feature extraction and recognition of vibration signals.Compared with the conventional two-dimensional convolutional neural network(2-D CNN)and one-dimensional convolutional neural network(1-D CNN),the proposed method balances the time and frequency scales well during feature extraction and reduces the number of pending parameters.A vibration recognition experiment is designed to classify three different vibration signals including knocking,damaging and interference.The recognition result shows that the MS 1-D CNN achieves good robustness to noise and similar recognition performance at twice processing speed compared to 2-D CNN.Thus,it is beneficial to improve the real-timing of vibration monitoring while maintaining the recognition performance.Based on the system and algorithm principle mentioned above,this paper builds a?-OTDR distributed optical fiber vibration signal monitoring interactive system,which integrates the functions of system control,parameter setting,alarm positioning,monitoring information record,vibration signal and map display,data saving and events replaying.The overall framework,monitoring process and the technical specifications of the hardware used of the system is described.Finally,the system monitoring performance is validated and analyzed via real experiments.