abstract
This paper presents the development of an approach for automatic estimation of the regression coefficients as a function of the refractive index (RI) in Tilted Fiber Bragg Gratings (TFBGs) based on their optical spectral characteristics. The methodology is based on machine learning approach, where different samples are analyzed in various test conditions to obtain a dataset for the training of a kNN regression algorithm. In this case, the spectral characteristics related to the wavelength spacing between adjacent cladding mode resonances, number of valleys and amplitude of the normalized spectra are correlated with the linear regression algorithm for RI estimation using the machine learning approach. The analysis are performed as a function of the wavelength shift and optical power, considering two regions on the transmitted spectrum and the envelope of the curve. Furthermore, the number of valleys and the area of the envelope curve are also analyzed as a function of the refractive index, which result in six analyzed features. The Principal Components Analysis (PCA) is used to reduce the number of features to two principal components. The results of the proposed algorithm are obtained with different trained and untrained samples, which indicates a root mean squared error (RMSE) of around 0.003 RIU for the trained samples and 0.008 RIU for the untrained ones. Therefore, the results indicate the usefulness of the proposed approach, which can be used to enhance the reproducibility of large set of TFBG sensors in practical applications.
keywords
SURFACE-PLASMON RESONANCE; OPTICAL-FIBER; SENSOR
subject category
Engineering; Optics; Telecommunications
authors
Leal, A Jr; Avellar, L; Frizera, A; Caucheteur, C; Marques, C
our authors
Carlos Marques
Assistant ProfessorProjects
Associated Laboratory CICECO-Aveiro Institute of Materials (LA/P/0006/2020)
acknowledgements
This research is financed by FAPES (458/2021 and 1004/2022), CNPq (310709/2021-0, 440064/2022-8 and 405336/2022-5), MCTI/FNDCT/FINEP (2784/20 and 0036/21) and Petrobras. This research is also financed by Fundacao para a Ciencia e a Tecnologia (FCT) through the 2021.00667.CEECIND (iAqua project) and PTDC/EEI-EEE/0415/2021 (DigiAqua project) . This work was developed within the scope of the project i3N (LA/P/0037/2020, UIDB/50025/2020 and UIDP/50025/2020) and CICECO (LA/P/0006/2020, UIDB/50011/2020, UIDP/50011/2020), financed by national funds through the FCT/MEC.