Research On Data Processing Method Of BOTDR Strain Detection System

In recent years,with the increasing of large-scale infrastructures in China,the demand for digital,real-time and early warning for structural health monitoring increases constantly.Brillouin 0ptical Time Domain Reflectometer(BOTDR)fiber optic sensing technology has been developing rapidly,and used in structural health monitoring of large infrastructures widely.However,due to the limitation of the measuring principle,the measuring time of traditional BOTDR system is long.The longer measuring distance means the longer the measuring time,thus it is difficult to realize fast monitoring.Aiming at the difficulty of fast monitoring in the traditional BOTDR system,A novel signal processing method in BOTDR strain measurement system is researched,using Graphics processing unit(GPU)parallel computing technology,and a fast parallel processing algorithm of BOTDR based on short time Fourier transform(STFT)is proposed.This method solves the problem of long timeconsuming in the traditional signal demodulation method,and finally achieve fast monitoring of strain.Aiming at large amount data,limited storage space and unability of remote monitoring,a distributed optical fiber strain remote monitoring system is researched and designed,based on data processing cloud,realizing digital remote monitoring of strain information and storage of massive data on cloud.In order to realizing of big data analysis in the future,large data supporting is provided by the data processing cloud.This paper mainly researches from the following aspects:(1)The strain detection principle of distributed fiber BOTDR strain detection system is researched.Firstly,various signal demodulation algorithms is analyzed in the current BOTDR system in detail,the implementation mechanism is studied deeply,and main signal demodulation algorithms are compared.After that,a fast parallel processing algorithm based on GPU parallel operation is proposed in STFT-BOTDR.(2)The GPU parallel computing hardware and software architecture and Compute Unified Device Architecture(CUDA)thread parallel mechanism is researched,the signal time-frequency conversion process of the STFT-BOTDR system is analyzed deeply.The STFT fast parallel processing algorithm based on GPU parallel computing is designed,the fast demodulation of BOTDR strain signal is realized.(3)The data processing cloud mechanism of the distributed optical fiber strain detection system is researched.The access interface of cloud database based on ADO technology is designed,and the upper computer program multi-thread concurrent access module is designed by using multi-thread mechanism,realizing fast and efficient data transmission.Finally,based on data processing cloud,a BOTDR strain remote monitoring database system is designed to store and analyze monitoring data,so does the collecting and processing of abnormal information,result in fast processing and remote monitoring of cloud data.(4)Through the above analysis,an experimental schemes is designed for the STFT-BOTDR fast parallel processing algorithm,and a software and hardware platforms is built.Experimental results show that,compared with the parallel algorithm of CPU platform,the GPU-STFT-BOTDR fast parallel processing algorithm has improved FFT time,accumulated average time and peak-finding algorithm execution time under the same parameter conditions.In the 10 km optical fiber,while the spatial resolution reaches 2m,1024 is the optimal parameter of FFT points,± 1.9MHz is the optimal measurement accuracy of brillouin frequency shift,the fastest demodulation speed of the system reaches 3 seconds.At last,the fast monitoring of strain can be realized.(5)Through the above analysis,an experimental schemes is designed for the distributed optical fiber strain remote monitoring system,the performance testing is carried out for the stability of the continuous operation of the test system,realtime responsiveness and the visual monitoring of database system.The testing results show that,this system can realize fast strain monitoring,cloud processing and cloud storage of abnormal data.This paper solve the problem of data processing and storage of large amount of data effectively in the digital management system.