Use of fixed wavelength Fibre-Bragg Grating (FBG) filters to capture time domain data from the distorted spectrum of an embedded FBG sensor to estimate strain with an Artificial Neural Network

Abstract

It is well known that an FBG sensor spectrum is distorted, thus making it difficult to estimate strain in monitored structures by simply tracking the peak point of the spectrum. Due to this issue, the traditional data acquired in the wavelength domain from optical spectrum analysers (OSA) needs significant processing to decode it into a useable form as an input to an Artificial Neural Network (ANN), a potential candidate for post-processing irregular data sets. This paper describes a successful application of an FBG filter system using three fixed wavelength FBG filters to capture real-time data from an embedded FBG sensor in the time domain. The time domain FBG data was post-processed using power-time area integration to account for the distortion before it was input into an ANN. The strain estimated by the ANN correlates well with the empirical results.