Research Of Distributed Optical Fiber Sensing Based On Rayleigh Scattering

Distributed optical fiber sensing is a kind of sensing technology which uses optical fiber to transmit information and can detect distributed information at various positions along the optical fiber.It has the characteristics of high pressure resistance,corrosion resistance and high measurement accuracy,and is widely used in the detection of various parameters.Coherent optical time domain reflectometer can not only measure temperature and strain,but also monitor disturbance signal.However,most of the current mainstream distributed optical fiber sensors can only realize the monitoring of a single parameter.If we want to measure multiple parameters at the same time,multiple systems need to be integrated,which is very costly.Therefore,the multi parameter detection and sensing technology is paid more and more attention.In view of the above defects,this thesis carries out in-depth research and analysis,and puts forward the solution of multi parameter monitoring of distributed optical fiber sensing technology,which overcomes some problems of industrial application of distributed optical fiber sensing technology.In this thesis,the sensing principle of coherent optical time domain is analyzed,and the numerical simulation of MATLAB is used to verify that the COTDR(coherent optical time domain reflector)system can realize the high-precision quantitative measurement of temperature and strain.In addition,the signal processing method is studied to denoise the signal,and the denoising effect of different methods is compared.Finally,the cumulative average denoising method is used to denoise the detected signal,so as to improve the accuracy of demodulation and lay the foundation for the follow-up research.A multi-function and multi parameter sensing design scheme based on coherent optical time domain reflectometer is proposed.According to the scheme,the specific selection of optical devices is given.Then the experimental system is simulated and the sensing experiment is studied.The sensing detection is realized in a 5km long optical fiber with 5m spatial resolution and 0.1?,1 ?? strain resolution of the system is improved.For the sample set and simulation signal set of optical fiber vibration signal in the field exploration on the network,the signal classification and recognition are studied by using the pattern recognition method.Firstly,the vibration signal is extracted by MATLAB and processed.Then,by comparing different signal features,the Mel cepstrum coefficient of the signal is used as the signal feature,which is used as the input of the neural network classifier for the training and testing of the classifier model.Finally,the pattern classification is carried out based on the behaviors of pedestrians,vehicles and environmental noise.Compared with the traditional time-frequency domain feature signals,the recognition accuracy of the final classifier is more than 95%.