Research On Signal Processing Technology Of SBS In Water For Temperature Measurement

With the development of spectral measurement technology, a novel method based on stimulated Brillouin scattering was proposed for measuring the water temperature in recent years. This method is a non-contact temperature measurement method which can be used to measure the temperature of water in real time. However, the measurement precision of water temperature is still an outstanding issue using the Brillouin method. In this paper, SBS signal was measured by using the pure water as scattering medium. The emphasis of the paper is focused on the signal processing techniques of SBS signal. Water temperature was accurately measured by measuring the frequency shift and line-width of SBS.The signal of SBS in the experiment was detected by the scanning system that made of F-P etalon and ICCD. The focus system was used to focus the pump light with 532 nm into a constant temperature tank so that the stimulated Brillouin scattering can be excited at the focus point. During the process of experiment, the ring spectra and dot spectra have been collected respectively, and then the obtained spectra have been processed so as to calculate the frequency shift and line width of SBS. Based on the processed results, the temperature values were calculated. The ring spectra were processed for obtaining the frequency shift of SBS based on two ways: one is the Lorenz-Gaussian fitting method; the other one is the nonlinear fitting method used to calculate the center position of the ring, and then the value of frequency shift can be obtained. The empirical formula was presented depending on the relationship between the frequency shift and temperature. The dot spectra were processed by Lorenz fitting method, and the empirical formula can also be given depending on the relationship between line width and temperature.Finally, the measurement error of three processing method have been analyzed. Conclusion can be drew from error analysis that the used Lorenz-Gaussian fitting method to require the value of frequency shift and temperature is simple and rapid, but the measurement error is relative larger(the maximum is 0.662℃, and the average is 0.279℃); the nonlinear fitting method for obtaining the value of frequency shift and temperature is more complex, and its maximum error is 0.420℃, the average error is 0.157℃. Although with the highest accuracy in theory, the Lorenz fitting method easily affected the signal while calculating the value of line width to get the temperature, and its maximum and average errors are 0.51 and 0.17 ℃ ℃ respectively.