Research On Multi-point Location Method And Implementation Of Distributed Optical Fiber Sensing System Based On ?-OTDR

The distributed optical fiber sensing system based on Phase-sensitive Optical Time Domain Reflectometer(?-OTDR)has the advantage of long distributed sensing distance,high sensitivity,strong anti-electromagnetic interference ability and simple structure.So,the system is widely used in oil and gas pipeline survey and security warning for important places,etc.In the actual application environment,the influence of noise and the signal intensity attenuation along with distance lead to the problem of low signal-tonoise ratio(SNR)of disturbance signal and inaccurately positioning in ?-OTDR system.Especially the far-end disturbance signal is weak and the SNR is very low,thus the accurate positioning of the far-end disturbance signal meets a serious challenge.Aiming at these problems,this thesis proposes a positioning scheme based on the combination of adaptive singular value decomposition denoising and improved K-means clustering.The adaptive singular value decomposition denoising method improves the SNR of the disturbance signal,and then the improved K-means clustering method is used to achieve accurate positioning of multi-point disturbances.The main research work achieved in the thesis is described as follows:(1)The basic theory of ?-OTDR and the principle of using this technology for disturbance detection are introduced.The possible noise sources in the ?-OTDR system are analyzed,which provides a theoretical reference for designing noise suppression schemes.(2)An airport perimeter security simulation experiment is set up,and the ?-OTDR system is used as a security warning system.There are four kinds of disturbance signals in the experiment,including watering,knocking,climbing and pressing.The single-point far-end disturbance experiment and the multi-point disturbance experiment are carried out respectively.According to the experimental data,empirical mode decomposition denoising,one-dimensional wavelet denoising,two-dimensional wavelet denoising and singular value decomposition denoising are studied respectively.In addition,the effect of the above denoising methods in removing Gaussian white noise is tested through simulation and comparison experiment.Numerical results show that,compared with the other three methods,the singular value decomposition method has the best denoising effect,can increase the SNR from 2.98 d B to 8.08 d B,and the time-consuming is the shortest,only 0.012 s.(3)The principle of particle swarm optimization algorithm is studied.Combining the particle swarm optimization and the singular value decomposition denoising method,an adaptive singular value decomposition denoising method is designed.This method can effectively improve the SNR of the disturbance signal.For four different types of disturbance signals,in the single-point far-end disturbance experiment,the SNR results are 9.55 d B,9.78 d B,12.29 d B and 10.44 d B;in the multi-point disturbance experiment,the highest SNRs are 11.38 d B,13.94 d B,10.80 d B and 12.19 d B,and the lowest SNRs are5.36 d B,6.95 d B,4.78 d B and 6.17 d B.The time consumption can be limited to less than3 s,thus the proposed method has good real-time performance.(4)An improved clustering method is proposed.Compared with the traditional Kmeans clustering method,the clustering method performs clustering in an ordered data set,thus greatly reducing the number of iterations and improving real-time performance.The experimental results show that the method can effectively locate the disturbance.In the single-point far-end disturbance experiment,the positioning accuracy rate is above95%.In the multi-point disturbance experiment,the positioning accuracy rate is above90%,and the average positioning accuracy rate reach 96.57%.The proposed method has great significance in promoting the wide application of ?-OTDR system.