Simultaneous Temperature Estimation of Multiple Gratings Using a Multi-Layer Neural Network

Abstract

This paper introduces a method to do simultaneous temperature estimations of multiple gratings in an FBG array. The method involves training a multi-layer neural network using simulated training data. The network is then used to estimate the temperature changes of multiple gratings simultaneously using only the experimentally obtained spectrum of the FBG array. The versatility of the method is seen from the results of three different setups. That is the broadband lightsource with OSA, the broadband lightsource with spectrometer, and incoherent optical frequency domain reflectometry. The method can estimate the temperature changes with high accuracy and low root mean squared error (RMSE) for the setups under consideration. Finally, the method is shown to be capable of simultaneous estimation of temperature changes for 19 FBGs using the BLS setup with a maximum absolute error of $6.64~\mathrm {\textbf {K}}$ and an RMSE of $1.69~\mathrm {\textbf {K}}$ in only 1.74 ms/FBG.