Application of a fiber Bragg grating temperature sensing method based on support vector regression optimized by a genetic algorithm for the decreasing external ambient temperature case

We studied the application of the fiber Bragg grating (FBG) temperature sensing method based on support vector regression optimized by a genetic algorithm (GA-SVR) for constant and decreasing external ambient temperature cases by simulation. The external ambient temperature could be retrieved from both the transient FBG wavelength and its corresponding change rate using GA-SVR, before the FBG temperature sensor reached the thermal equilibrium state with the external ambient temperature. FBG wavelengths and their corresponding change rates in the cases of FBG sensor temperatures higher and lower than the external ambient temperature were studied and used to construct the training data set. We found that there exist singularity points in the curves of the wavelength change rate when the FBG sensor temperature is higher than the external ambient temperature in some cases, which is different from the case where the FBG sensor temperature is lower than the external ambient temperature. Its application for sensing the constant and decreasing external ambient temperature in real time was demonstrated with an accuracy of 0.32°C in those two cases. It also indicates that for real applications of this temperature sensing method where the external ambient temperature varies randomly, the FBG sensor temperature changes rather than the external ambient temperature changes play the dominant role. What is more, the demodulation time was decreased to 0.002 s, which is approximately 0.05‱ of the time constant of the FBG temperature sensor. In other words, this method makes it possible to realize the external ambient temperature determination using a time smaller than the time constant of the FBG sensor. The high sensing accuracy and fast demodulation speed are crucial for future high-performance real-time FBG temperature sensing.