Design Of Distributed Raman Temperature Measurement System Based On GPU

As the speed of industrial upgrading in China continues to accelerate,the corresponding safety testing measures have also been upgraded.Temperature detection is an important part of safety monitoring.Among them,the distributed optical fiber temperature sensing system based on the principle of Raman optical time domain reflection(R-OTDR)can monitor the temperature signal.However,as the monitoring distance becomes longer and the monitoring range expands,the real-time performance of its distributed optical fiber temperature sensor becomes an urgent problem to be solved.Distributed Fiber Temperature Sensing Detection Technology using multimode fiber as the sensing unit and transmission medium is now proposed.Aiming at the real-time problems of this technology in long-distance temperature monitoring,it is proposed to use the advantages of Graphics Processing Unit(GPU)parallel computing in data processing to accelerate the system,shorten the long-term temperature detection time of the system,and solve the existing realtime problems of ROTDR system.Develop a GPU-based R-OTDR temperature sensor,which has the advantages of wide measurement range,high system stability,intrinsic safety,strong resistance to corrosion and interference.The research work in this paper mainly includes the following aspects:(1)This paper researches on the temperature measurement principle and signal denoising analysis of the distributed temperature sensing system.The factors that affect the signal-to-noise ratio of the system signal in temperature signal acquisition are studied and analyzed.In order to ensure the accuracy of the acquired signal,a combined denoising method using a cumulative average and wavelet threshold is proposed to improve the signal-to-noise ratio of the temperature signal feasibility.(2)This paper analyzes the theory of the GPU architecture and the principles of its parallel computing,by using its advantages in parallel computing to design a fast distributed optical fiber temperature sensing system scheme combined with GPU.According to the system plan and the technical objectives of the subject research,the parameters of each key module are analyzed,and the experimental platform is built.(3)According to the characteristics of the Compute Unified Device Architecture(CUDA)architecture,we using a CUDA C language to programming denoising algorithms based on the GPU hardware platform,and write the upper computer software in conjunction with Microsoft Foundation Classes(MFC).According to the system functional requirements,the software has functions such as system startup control,temperature signal reading and processing,temperature signal demodulation and reality,and data storage.(4)Combining the experimental hardware platform and the host computer software system,the prototype integration of the distributed optical fiber temperature sensing system is completed.Different experiments were designed to test the performance of the distributed Raman optical fiber temperature sensor system.Within a measurement distance of 10 km and a temperature measurement range of-25℃～120℃,measurement accuracy of ±0.6 ℃ was achieved,and the positioning accuracy reached 2m Measuring time 3.6s.Compared with the ordinary distributed optical fiber temperature sensor,it has higher measurement accuracy and shorter measurement time.