Machine Learning Assisted Fiber Bragg Grating-Based Temperature Sensing

This letter proposes an alternative approach to the signal processing of temperature measurements based on fiber Bragg gratings (FBGs) using the machine learning tool Gaussian process regression (GPR). The experimental results show that for a majority of the cases under consideration, the reported technique provides a more accurate calculation of the temperature than the conventional methods. Furthermore, the GPR can give the uncertainty of an estimate together with the estimate itself, which for example is useful when it is important to know the worst-case scenario of a measurand. The GPR also has the potential to improve the measurement speed of FBG-based temperature sensing compared to current standards.