Signal Processing And Performance Improvement For Distributed Fiber Raman Temperature Sensor

Due to continuous measurement, as well as small size, resistant to radiation and strong stability, the distributed optical fiber temperature sensors have been a unique technology in the ptical fiber sensing technology. With more and more people’s attention, this technology has a significant progress. In distributed fiber sensors, the distributed fiber Raman temperature sensor is most mature and convenient, In the distributed fiber Raman temperature sensor, the temperature is obtained by the backward Raman scattering signal, and the fiber position is determined by the Optical time domain reflection technology (OTDR). Based on the backward Raman scattering signal and OTDR, the continuous temperature along the whole sensing fiber can be obtained. Beacause the traditional temperature sensors are installed in some fixed points, so these sensors can not be applied widely in these continouse temperatature distribution areas. The traditional sensors not only have high high installation, but also are difficult to install. On the contrary, the distributed fiber Raman temperature sensor the continuous temperature along the whole sensing fiber can be obtained, In these sensors, the sensing fiber not only senses the temperature, but also transmits temperature information. In that, these sensors have less cost. In distributed fiber sensors, the distributed fiber Raman temperature sensor is based on backward Raman scattering signal and weak signal detection technology and it could be applied in hazardous conditions owing to its anti-electromagnetic interference, good insulativity,high sensitivity and high real-time. It has been applied in many fields, such as: Industry, construction, aerospace, petroleum, electric power and military and other fields.In the distributed fiber Raman temperature sensor, the temperature is obtained by the backward An-Stokes and Stokes scattering signal, and the fiber position is determined by the Optical time domain reflection technology (OTDR). Based on the backward Raman scattering signal and OTDR, the continuous temperature along the whole sensing fiber can be obtained. Due to weak backward scattering signal, the Signal to Noise Ratio(SNR) will be less if the signal is without signal processing. In order to improve system performace, the signals have to be processed by the proper method. So the study on signal processing is important to improve SNR and system performance. Beyond signal processing method, the reasonable system design playes an important role in improving system index. Because the reasonable system design can eliminate the influcece of external environment effectively, such as: ambient temperature and disturbance, so it has an important role in improve to system stability.With the development of science and technology, people have lager requirement on the performance of distributed fiber Raman temperature sensor. So the study on signal processing and the optimal design of system are of great importance to improve system performance and meet the people’s demand.In this paper, the design, the analysis of operating principle and the research of system noise and performance are conducted. Base on the research proposed above, the signal processing method and optimizing design scheme were introduced. The main contents and innovations are as follows:(1) The theory of optical scattering was described including Rayleigh scattering, Brillouin scattering and Raman scattering. The Raman scattering theory was analyzed from traditional electromagnetic theory and quantum theory specially. Moreover, the temperature features of the three scatterings were introduced. According to the temperature features, the system principle and structure of the distributed optical fiber temperature sensors based on Raman and Brillouin scatterings were described, and also the principle of optical time domain reflectometry. At last, the demodulation methods of the distributed Raman optical fiber temperature sensor were analyzed and investigated, including: single-path demodulation based on anti-Stokes, dual-paths demodulation based on anti-Stokes and Rayleigh and dual-paths demodulation based on anti-Stokes and Stokes.(2) Based on dual-paths demodulation based on anti-Stokes and Stokes, the design scheme of distributed Raman optical fiber temperature sensor was introduced, including the instrument parameters, function and the effect on the system performance. Then the main system performance indices were describled, involving temperature precision, temperature resolution, spatial resolution, positioning accuracy, time resolution and sensing distance.Moerover, the influnce factors of these indices were analyzed. At last, the noise source was analyzed comprehensively, and it makes an important significance on the improving system performance.(3) The signal processing methods of distributed Raman temperature sensor was introduced, including cumulative average algorithm and wavelet transform modulus maximum. The principles of the two methods were decaribled respectively. In the cumulative average algorithm, the denoising effect of the cumulative numbers is analyzed by the experiments. The principle of wavelet transform and the signal processing methods based on wavelet transform were introduced respectively. The principles of wavelet decomposition and reconstruction method, the nonlinear wavelet threshold method and wavelet transform modulus maximum are descibled. In the wavelet transform modulus maximum, the realization method of wavelet modulus maxima, the determination method of wavelet decomposition scale and the method of choosing wavelet modulus maxima signal are introduced respectively. Moreover, based on the low SNR, an improved wavelet transform modulus maxima (WTMM) was proposed and demonstrated. In this scheme, the WTMM were obtained by combining those on the high and low decomposition scales.(4) A theoretical model to analyze the impact of Rayleigh noise on Raman distributed temperature sensors, which use the anti-Stokes and Stokes light or anti-Stokes component only as the demodulation signals was presented. Based on this model, the effects of Rayleigh noise on temperature accuracy, sensitivity, and resolution are investigated both at only one point and in a section of the fiber. In order to improve system performance, two methods were proposed. A method utilizes a reference fiber with different temperatures and sensing fiber with a reference temperature and a measuring temperature, such that the Rayleigh scattering is eliminated by the difference between the two signals at different reference tempearatures of reference fiber, and the difference between signals at reference temperature and measuring temperature for sensing fiber. In other method, the Rayleigh noise can be calculated and then eliminated simultaneously by the intensity of the signal composed by anti-Stokes light and Rayleigh noise at two different temperatures of two reference fibers.(5) The factors that fiber attenuation variation influenced by the environment and the error caused by the attenuation calibration affect the stability of the system. According the problem, a novel method to auto-correct and compensate the attenuation in Raman distributed temperature sensors is proposed. This method utilizes a fiber in loop configuration combined with a reference temperature in the front section to cancel out the impact of attenuation generated by different wavelengths, local external perturbations and environment changes.(6) Ensuring the synchronization of backscattered Raman light and improving the sampling rate are of great importance to improve the spatial resolution of a distributed Raman temperature measurement system. The asynchrony of backscattered Raman light is mainly caused by the hardware and dispersion. To keep the synchronization of the Anti-Stokes and Stokes light, three effective methods are proposed in this paper. First, the asynchrony caused by the difference in the hardware is eliminated by attaching a fiber to the pigtail of APD. Next, the data acquisition card chooses different sampling rates for the Anti-Stokes and Stokes light according to the velocity of propagation. Then, a dispersion auto-correction algorithm is adopted to decrease the difference in the fiber position between the Anti-Stokes and Stokes signals. Finally, the optical switch with different length of pigtails is applied to improve the sampling rate without upgrading the hardware of data acquisition card.