Research On High Speed Demodulation System Of Fbg Sensor Based On FPGA And GPU

As one of the most promising passive optical devices,fiber Bragg grating(FBG)has attracted worldwide attention since it was first used in temperature and strain sensing by Morey.With the continuous progress of grating writing technology and the emergence of ultra-low reflectivity weak fiber grating,the multiplexing ability of FBG has been greatly improved,which makes the FBG sensor network develop towards the direction of large-scale,long-distance and high-precision.How to demodulate the data of large-scale optical fiber sensor network accurately,quickly and cheaply to meet the purpose of real-time online monitoring has become a new focus of researchers.This dissertation makes use of the advantages of field programmable gate array(FPGA),such as flexible design,parallel pipeline,rich memory interface and excellent logic control,and graphic processing unit,GPU)has powerful floating-point and parallel computing capabilities.A fiber Bragg grating sensor system based on FPGA and GPU high-speed demodulation is designed and developed,which realizes the detection of weak grating wavelength under continuous output laser scanning.Because of the increase of demodulation speed,the dynamic monitoring range of the system is improved.It can not only detect the static physical quantities such as temperature and strain,but also detect the low-frequency vibration signal.The main contents of this dissertation are summarized as follows:(1)Based on the basic theory of FBG,the detection principle of FBG sensing system is described,and the demodulation methods of several typical FBG sensing systems are studied and compared.The principle of large-scale FBG sensor multiplexing based on TDM is analyzed in detail,and the influence of multiple reflection crosstalk and shadow effect on the multiplexing ability of the system is analyzed.In the wavelength demodulation scheme based on the tunable scanning method of light source wavelength,the Gauss fitting algorithm of peak seeking algorithm is proposed to demodulate the peak position of reflection spectrum and improve the demodulation accuracy of the system.(2)On the basis of theoretical analysis,a quasi distributed FBG sensing system based on wavelength demodulation is constructed,and the sensing principle of the system is described in detail.The system's scanning light source module,semiconductor optical amplifier(SOA)pulse modulation module and other hardware modules are described.From the four aspects of acquisition,storage,transmission and demodulation,the design of high-speed demodulation system for FBG sensor system is completed.Data acquisition: a high-speed ADC chip with 250 m sampling rate and double data rate(DDR)data transmission mode is adopted.Data storage: DDR3 with large capacity and ultra-high speed data reading and writing is used as the main data cache space,and ping-pong operation data flow mode is used.Data transmission: Based on the high-speed serial computer expansion bus standard(PCIe),use Riffa architecture to accelerate it.Data demodulation: for data demodulation,a time-saving parallel computing method is adopted,and demodulation calculation is completed under Compute Unified Device Architecture(CUDA)of GPU.(3)Set up a wavelength demodulation system in the laboratory.The correctness,stability and dynamic measurement range of the system are verified by experiments.In order to verify the actual effect of using GPU to speed up demodulation,the peak seeking algorithm of Gaussian fitting is implemented in CPU,FPGA and GPU respectively,the demodulation time is counted,and the comparative test is completed.