Demodulation Method of Tilted Fiber Bragg Grating Refractometer Based on Gramian Angle Field Algorithm and 2D Residual Convolutional Neural Networks

Abstract

Tilted fiber Bragg grating (TFBG) has been widely for refractive index sensing due to their high sensitivity, but there is a lack of demodulation method with high accuracy and good real-time performance. In this paper, a novel demodulation method based on deep learning is proposed and verified experimentally. As demonstration, a bare TFBG working as refractometer was immersed in sodium chloride solution with different concentrations, whose RI ranged from 1.3350 to 1.3790 RIU. The dimension of the acquired transmission spectra was ascended by the Gramian Angle Field (GAF) algorithm and then demodulated by a 2D residual convolutional neural networks (CNN). As a result, the accuracy represented by the coefficient of determination reaches 99.75% and the mean square error (MSE) is only 4.049 × 10-7. The proposed method makes it possible to achieve a fast, robust, and accurate demodulation TFBG refractive index sensing scheme.